catalytic transformation of ethanol to 1,3-butadiene over mgo/sio2 catalyst
TRANSCRIPT
Lehigh UniversityLehigh Preserve
Theses and Dissertations
2018
Catalytic Transformation of Ethanol to1,3-Butadiene over MgO/SiO2 CatalystWilliam E. TaifanLehigh University
Follow this and additional works at: https://preserve.lehigh.edu/etd
Part of the Chemical Engineering Commons
This Dissertation is brought to you for free and open access by Lehigh Preserve. It has been accepted for inclusion in Theses and Dissertations by anauthorized administrator of Lehigh Preserve. For more information, please contact [email protected].
Recommended CitationTaifan, William E., "Catalytic Transformation of Ethanol to 1,3-Butadiene over MgO/SiO2 Catalyst" (2018). Theses and Dissertations.4255.https://preserve.lehigh.edu/etd/4255
Catalytic Transformation of Ethanol to
1,3-Butadiene over MgO/SiO2 Catalyst
by
William E. Taifan
A Dissertation
Presented to the Graduate and Research Committee
of Lehigh University
in Candidacy for the Degree of
Doctor of Philosophy
in
Chemical Engineering
Lehigh University
May 2018
ii
Copyright © 2018 by William E. Taifan
iii
CERTIFICATE OF APPROVAL
Approved and recommended for acceptance as a dissertation in partial fulfillment
of the requirements for the degree of Doctor of Philosophy.
Date
Accepted Date
Dissertation Director:
Date Dr. Jonas Baltrusaitis, Ph.D.
Advisor and Committee Chairperson
Assistant Professor of Chemical
Engineering
Lehigh University
Committee Members:
Date Dr. Israel E. Wachs, Ph.D.
G. Whitney Snyder Professor of
Chemical Engineering
Lehigh University
Date Dr. Hugo Caram, Ph.D.
Professor of Chemical Engineering
Lehigh University
Date Dr. Nicholas C. Strandwitz, Ph.D.
Assistant Professor of Materials
Science and Engineering
Lehigh University
iv
ACKNOWLEDGEMENTS
This dissertation is made possible by the experiences I have had during my graduate school,
where I have had encounters with wonderful individuals that have been impactful in their
own rights. I am dedicating this page to demonstrate my gratefulness for the
encouragement, stimulation, and motivation that I had from these individuals.
I am eternally grateful for my family’s support for my PhD endeavor. Unsurpassed love,
passion, and attention from both my mother and father have been a driving force to
overcome the potential (mental) barrier at difficult times during my research days. I am
thankful that, if not for my father, I would not have even chosen chemical engineering as
my major. I am thankful that, if not for him, I would not have pursued my doctoral degree.
While my father had been the light that I follow, my mother had been my best supporters
throughout my graduate study. My brother has played an instrumental part in my life, as
we share a lot of secrets that are impossible to share with our parents. These secret-sharing
sessions had been accompanied by advices sharing, where giving him advices had made
me a more mature, better human being. I am grateful of my grandmother and grandfather
for their unwavering support, and for their nagging me to find a wife for myself. Since the
wife is finally here, you rest assured that your first grandson will give you some great
grandchildren. I love you all, and I can never envision completing my PhD without the
support from all of you.
Lehigh has not only given me an opportunity for a graduate study, instead, it has given me
my life partner. The encounter we had in Taylor Gym made my dream of having a spouse
with similar passion came true. If there was a little bright side of living in Lehigh, that
would be you. You are my first in a lot of aspects, such as being a medical emergency
contact during your surgery and taking care of your leg. Those experiences changed me as
a person, for the better. I love you very much, so much that I had decided to commit a
lifetime’s worth of opportunities to experience all the steakhouses and sushi places in Japan
and New York City.
To all of my collaborators. Tomáš Bučko (forgive me if I mistyped the accent on your
name), my first ever collaborator, I can’t imagine how my first ever manuscript would be
possible without your help. I gained a lot of insights on DFT calculation, including few
tricks to stabilize those explosive transition states. Dr. Yuanyuan Li and Prof. Anatoly
Frenkel from Brookhaven National Lab, I’m eternally grateful for giving me a free ride on
your XAS proposal to finally nail down the active sites on my catalyst. The sleepless nights
we spent during Thanksgiving break (yes, during that break), finally paid off and hopefully
the manusript got accepted without a lot of revisions. I’m also acknowledging Dr. Nebojsa
Marinkovic, Nicholas Marcella, Amani Ebrahim, for your help in the beamline.
To my lifting dudes Tony Chang, Henry Choo, Steven Rodriguez, I’m grateful for the
knowledge we share during our lifting sessions. From Candito to Sheiko, from percentages
to RPE. I’m grateful for our meet together, it was a very profound experience. Powerlifting
v
had been a part of my graduate school life. My love for this sport is a causal relationship
with my research, where the former has been a therapy to the detrimental effects of the
latter. I’m grateful as well that this sport has introduced me to a bunch of wonderful people,
Shane Del Bianco, Johnny Luczkovich, Mark Herndon, and my coach, Samuel Bernstein.
May we all be relentless in our pursuit of strength and power. May our backs arch
beautifully during our bench, may we squat lower than our GPA, and may our deadlifts be
three times our bodyweight.
During the early dark days of the research I have also taken solace in my friendships with
great individuals at this university. With no particular order: George Xu Yan, Benjamin
Moskowitz, Fan Ni, Aaron Zhang, Daniyal Kiyani, Chris Keturakis, Yoona Yang, Leah
Spangler, Sagar Sourav, Chris Curran, Evan Koufos, and Lohit Sharma. My mentors from
Wachs’ lab: Minghui Zhu, Soe Lwin, Jih-Mirn Jeng, and Ivan Santos. Please accept my
apologies for missing your names in this list, as you see I have made efforts to include
everyone’s names, but you all are very meaningful to me.
Lastly I would also like to thank the two Professors that are very influential to me during
my PhD study. Professor Israel E. Wachs, for the meaningful discussion and wonderful
lectures on the operando catalyst characterization, emphasizing the importance of under
reaction condition characterization. I gained a lot of knowledge from our discussions
during the lecture and about my PhD work. My adviser, Jonas Baltrusaitis, with whom I
have formed a long-lasting relationship, though I’m a bit bitter that he couldn’t hood me
on my graduation day. He had given me opportunities to travel and attend the biggest
conferences throughout my PhD study. From him I acquired my knowledge, attitude, and
mindset of a hybrid engineer-scientist. I have grown much since I first came here in terms
of both personal and professional development. I am eternally grateful for your advises and
guidance throughout my graduate study, as they all said, you only have one Ph.D. advisor
for the rest of your life.
vi
TABLE OF CONTENTS
ACKNOWLEDGEMENTS............................................................................................ iv
TABLES OF CONTENTS ............................................................................................. vi
LIST OF TABLES .......................................................................................................... x
LIST OF FIGURES .......................................................................................................xii
ABSTRACT ..................................................................................................................... 1
CHAPTER 1 | Introduction
1. Background .......................................................................................................... 4
1.1. ETB Reaction Network ............................................................................. 10
1.1.1. Reaction Intermediates and Byproducts ................................................... 11
1.1.2. Proposed Reaction Mechanisms ............................................................... 12
1.2. Catalytic Systems ..................................................................................... 15
1.2.1. Reaction Conditions and Catalytic Performance ...................................... 16
1.2.2. MgO/SiO2 Catalysts ................................................................................. 18
1.2.3. ZrOx –based Catalysts ............................................................................... 24
1.2.4. Other Catalysts ......................................................................................... 26
2. Approach ............................................................................................................ 29
2.1. Approach .................................................................................................. 29
2.2. DFT Calculation ....................................................................................... 30
2.3. In-situ and Operando Spectroscopy ......................................................... 32
2.3.1. Infrared Spectroscopy ............................................................................... 33
2.3.2. UV-Vis Spectroscopy ............................................................................... 34
2.3.3. Operando XANES and EXAFS ................................................................ 35
2.3.4. Temperature-Programmed Reaction Spectroscopy .................................. 36
2.4. HS-LEIS ................................................................................................... 37
2.5. Probe Molecules ....................................................................................... 38
2.6. Product Determination with GC-MS ........................................................ 39 3. Thesis Outline ..................................................................................................... 41
References ....................................................................................................................... 43
CHAPTER 2 | Experimental Methods
1. Introduction ........................................................................................................ 48
2. Computational Details ....................................................................................... 49
2.1. Electronic structure calculations .............................................................. 49
2.2. Structural optimization calculations ......................................................... 49
2.3. Structural model ....................................................................................... 50
2.4. Free-energy calculation ............................................................................ 52
i. Working equations ................................................................................... 52
ii. Partitioning of atomic degrees of freedom in interacting and non-interacting
systems ................................................................................................................. 53
iii. Calculation of harmonic vibrational frequencies ..................................... 54
vii
3. Experimental Methods ...................................................................................... 55
3.1. Catalyst synthesis ....................................................................................... 55
3.1.1. Synthesis of magnesium oxide, MgO, catalyst ......................................... 55 3.1.2. Synthesis of MgO/SiO2 catalysts .............................................................. 55
3.1.3. Synthesis of promoted wet-kneaded MgO/SiO2 catalysts ......................... 56
3.2. Catalytic reactivity study ........................................................................... 56
3.3. Catalyst characterization ............................................................................ 58
3.3.1. High-sensitivity low energy ion scattering (HS-LEIS) .............................. 58
3.3.2. XRD and BET surface area ........................................................................59
3.3.3. Transition metal concentration measurements ........................................... 59
3.3.4. Scanning transmission electron microscopy .............................................. 60
3.3.5. In-situ spectroscopy ................................................................................... 60
3.3.6. Acid-base characterization using pyridine, NH3, CO2, and methanol as
probe molecules .................................................................................................... 61
3.4. Reaction mechanism study using in-situ DRIFTS spectroscopy and TPRS
............................................................................................................................... 62
3.5. Operando XANES and EXAFS spectroscopy during ethanol reaction to
1,3-BD over Cu- and Zn-promoted MgO/SiO2 catalysts ...................................... 64
References ....................................................................................................................... 65
CHAPTER 3 | Computational Study of Ethanol to 1,3-BD
Reaction Mechanisms
Abstract ........................................................................................................................... 68
1. Introduction ........................................................................................................ 69
2. Computational Results ....................................................................................... 73
2.1. Reaction Pathways .................................................................................. 73
2.1.1. Ethanol dehydrogenation and dehydration ............................................. 73
2.1.2. Aldol condensation ................................................................................. 80
2.1.3. Prins condensation .................................................................................. 85
2.1.4. 1-Ethoxyethanol formation ..................................................................... 88
2.2. Details of the free-energy profiles .......................................................... 89
2.2.1. Elimination/redox reaction of ethanol .................................................... 90
2.2.2. C-C bond formation ................................................................................ 90
2.2.3. Proton transfer ......................................................................................... 91
3. Discussion ............................................................................................................ 92
4. Conclusion ........................................................................................................... 98
Supporting Information .............................................................................................. 100
References ..................................................................................................................... 112
CHAPTER 4 | Combined In-situ DRIFTS and DFT study of
Ethanol to 1,3-BD Reaction Mechanism over MgO/SiO2 catalysts
Abstract .......................................................................................................................... 103
1. Introduction ....................................................................................................... 104
viii
2. Results and Discussion ...................................................................................... 109
2.1. Catalyst activity and selectivity testing ....................................................... 109
2.2. In-situ DRIFT spectroscopy of MgO based catalyst surface hydroxyl
groups................................................................................................................... 110
2.3. Acid-base characterization of WK (1:1) catalyst using CO2 and pyridine
as probe molecules .............................................................................................. 113
2.4. In-situ DRIFT spectroscopy to monitor hydroxyl group reactivity during
the ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol adsorption and
subsequent reaction on a WK (1:1) catalyst surface ........................................... 115
2.5. In-situ DRIFT spectroscopy of C2 (ethanol, acetaldehyde) and C4
(crotonaldehyde and crotyl alcohol) adsorption and reaction on WK (1:1)
catalyst surface as a function of temperature ...................................................... 119
2.5.1. C2 reactants and intermediates ............................................................... 119
2.5.2. C4 intermediates ..................................................................................... 127
2.6. DFT calculations ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol
vibrational frequencies ........................................................................................ 130
2.7. In-situ DRIFT spectra for the ethanol, acetaldehyde, crotonaldehyde and
crotyl alcohol reaction on a WK (1:1) catalyst surface: the effect of the vapor
phase presence .................................................................................................... 136
3. Conclusions ........................................................................................................ 145
Supporting Information ............................................................................................... 148
References ...................................................................................................................... 151
CHAPTER 5 | Active Sites Determination of MgO/SiO2 Catalysts
for Ethanol to 1,3-BD Reaction
Abstract .......................................................................................................................... 154
1. Introduction ....................................................................................................... 155
2. Results and Discussion ...................................................................................... 160
2.1. Steady state ethanol catalytic conversion to 1,3-BD ................................. 160
2.2. Bulk, surface chemical and structural characterization using XRD, LEIS
and DRIFTS ........................................................................................................ 163
2.3. Temperature-programmed reaction spectroscopy (TPRS) of ethanol on
MgSi-WK ............................................................................................................ 167
2.4. Acid-base characterization using DRIFTS ................................................ 172
2.5. Reactive site persistence during ethanol-to-1,3-BD .................................. 174
2.6. Implications for the structure-activity relationship .................................... 186
3. Conclusions ........................................................................................................ 189
References ...................................................................................................................... 191
CHAPTER 6 | Role of transition metal promoters (Cu, Zn) on
MgO/SiO2 catalyst for Lebedev reaction
Abstract ..........................................................................................................................194
1. Introduction .......................................................................................................195
ix
2. Computational results ...................................................................................... 199 2.1. Model catalyst selection and analysis ...................................................... 199
2.2. Reactive intermediates ............................................................................. 206
2.3. Potential energy surfaces .......................................................................... 210
3. Experimental results .........................................................................................214 3.1. Catalyst characterization .......................................................................... 214
3.2. Steady state catalytic performance and acid/base chemistry of the
catalyst active sites .............................................................................................. 222
3.3. Active sites under operating conditions ................................................... 226
3.3.1. Temperature programmed infrared spectroscopy measurements (TP-
DRIFTS) ............................................................................................................. 226
3.3.2. In-situ UV-Vis DRS study of MgSi catalysts . 230
3.3.3. Operando XAS studies of Cu, Zn-promoted MgSi catalysts ................... 232
3.3.3.1. Operando XANES and EXAFS of Cu-promoted MgSi catalyst ......... 232
3.3.3.2. Operando XANES and EXAFS of Zn-promoted MgSi catalyst ......... 241
4. Conclusion ......................................................................................................... 245
Supporting Information ............................................................................................... 247
References ...................................................................................................................... 266
CHAPTER 7 | Conclusions and Future Outlook
1. Conclusions......................................................................................................... 270
2. Future Outlook................................................................................................... 273
References ......................................................................................................................275
CURRICULUM VITAE ................................................................................ 276
x
LIST OF TABLES
Chapter 1
Table 1.1. Catalytic performance of different catalysts studied for one-step ethanol to
1,3-BD conversion. a Contact (residence) time; WHSV (weighted-hourly space
velocity); TOS (time-on-stream); X (ethanol conversion), Y (yield); P (productivity)
......................................................................................................................................... 17
Chapter 3
Table 3.1 Electronic and free energy values of the stationary points calculated at 0 K
and 723 K, respectively .................................................................................................. 75
Table 3.2 Computed forward and reverse reaction barriers and the corresponding
reaction rate constants ..................................................................................................... 95
Chapter 4
Table 4.1. Catalytic activity comparison of WK (1:1) with previously investigated wet-
kneaded synthesized catalysts ......................................................................................... 109
Table 4.2. Surface hydroxyl group vibrational frequencies during ethanol,
acetaldehyde, crotonaldehyde and crotyl alcohol adsorption on WK (1:1) ................... 116
Table 4.3. Vibrational frequencies and their assignments for ethanol, acetaldehyde,
crotonaldehyde and crotyl alcohol adsorption on WK (1:1) ........................................... 121
Table 4.4. Calculated infrared frequencies of ethanol, acetaldehyde, crotonaldehyde
and crotyl alcohol molecules adsorbed on low coordination model MgO surface sites.
Frequencies were calculated using PBE density functional and no scaling to correct for
anharmonicity was applied ............................................................................................. 135
Table S4.1. Calculated infrared frequencies of gas phase ethanol, acetaldehyde, crotyl
alcohol and crotonaldehyde molecules. Frequencies were calculated using PBE density
functional and no scaling to correct for anharmonicity was applied. Experimental
frequencies, except for crotonaldehyde, were obtained from NIST .............................. 149
Chapter 5
Table 5.1 Steady state reactivity of MgO/SiO2 catalysts of different calcination
temperature and preparation method. Reaction was carried out at 450 °C with catalyst
mass of 0.1 g, 55 ml/min total flow rate and pethanol = 2.5 kPa. Selectivity towards major
(by)products ethylene, acetaldehyde and 1,3-BD is reported ......................................... 161
Table 5.2 m/z selection to identify the arising vapor-phase species from TPRS
experiments...................................................................................................................... 168
xi
Table 5.3. Comparison between observed experimental values of NH3 adsorption on
MgSi-WK catalysts with DFT calculated IR vibrations of NH3 adsorbed on open and
closed acid Mg3C and Mg4C sites. Scaling factor of 0.9854 was applied to the calculated
values and was derived from the gas-phase NH3 experimental and DFT calculated
frequencies....................................................................................................................... 182
Chapter 6
Table 6.1 Different configurations tested for Zn(Cu)/MgO model catalysts. Various
dopant location was chosen between the top and second layer, and compared for energy
and Bader charge. ............................................................................................................203
Table 6.2. Referenced electronic and corrected Gibbs free energy for each species over
MgO, Cu/MgO, and Zn/MgO catalysts. ......................................................................... 208
Table 6.3 Activation energy and thermodynamics consideration for key steps during
ethanol conversion to 1,3-butadiene over MgO, Zn/MgO, and Cu/MgO catalysts......... 211
Table 6.4. Vibrational frequencies in 1600-1400 cm-1 wavenumber range and their
assignments for ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol adsorption
on WK (1:1)11.................................................................................................................. 230
Table 6.5. Best fitting results of Cu catalysts. The structural parameters of standards
were listed for comparison............................................................................................... 233
Table 6.6. Best fitting results for ZnMgSi, ZnMg, ZnO, MgO and Zn. The structural
parameters of standards were listed for comparison. ...................................................... 245
Table S6.1. Peak assignments of surface CO2 species identified on MgSi, CuMgSi, and
ZnMgSi catalysts............................................................................................................. 251
Table S6.2. DFT simulation of NH3 on MgO slab. Simulation was done using VASP,
PBE functionals on 2x2x1 k-point mesh......................................................................... 256
Table S6.3. Redox properties of the MgSi, CuMgSi and ZnMgSi catalysts and
reference MgO obtained from MS measurements. These results have been normalized
to the BET surface area (m2/g) of each catalysts ............................................................ 263
xii
LIST OF FIGURES
Chapter 1
Figure 1.1. Ethanol production rate increase from 2010 to 2017 (adapted from U.S.
Energy Information and Administration) (left) and ethanol upgrading map to different
highly valued chemicals (right)....................................................................................... 4
Figure 1.2. Proposed reaction mechanisms for ethanol conversion to 1,3-BD: (a)
Toussaint’s aldol condensation; (b) Gruver’s Prins condensation; (c) Cavani’s
carbanion mechanism ......................................................................................................13
Figure 1.3. Operando spectroscopy setup, flow reaction cell temperature/pressure
controller equipped with FTIR, UV-Vis and Raman spectroscopy that enables real-time
online measurement. Output is connected to real-time GC/MS system. Adapted from: http://www.lehigh.edu/operando......................................................................................
. 33
Chapter 2
Figure 2.1 Periodic MgO slab used throughout the calculations. The whited out bottom
layer indicates the atoms whose positions were kept frozen during
calculations...................................................................................................................... 51
Chapter 3
Figure 3.1. Reaction mechanisms proposed for ethanol to 1,3-butadiene; (a)
Toussaint’s generally accepted mechanism, (b) Fripiat’s Prins mechanism, (c)
Ostromislensky’s hemiacetal rearrangement................................................................... 72
Figure 3.2 All stable intermediates and transition states calculated following the
reaction pathways. (1A-1C): ethanol dehydrogenation to acetaldehyde; (2A-2O):
acetaldehyde aldol condensation to 3-hydroxybutanal (acetaldol) followed by proton
transfer to crotonaldehyde; (3A-3G): MPV (Meerwein–Ponndorf–Verley) reduction of
crotonaldehyde to 1,3-butadiene; (4A-4K): acetaldol MPV reduction to butadiene; (5A-
5C): ethanol dehydration to ethylene; (6A-6E iii 3): Prins condensation of acetaldehyde
and ethylene; (7A-7E): ethanol and acetaldehyde nucleophilic addition reaction
(Ostromislensky’s hemiacetal rearrangement)................................................................ 79
Figure 3.3 Free-energy profiles for (a) ethanol dehydrogenation to form acetaldehyde
and (b) ethanol dehydration to ethylene...........................................................................79
Figure 3.4 Free-energy profiles for aldol condensation pathway................................... 81
Figure 3.5 Free-energy profiles for the MPV reduction of the resulting molecule from
aldol condensation. Red pathway indicates subsequent proton transfer of acetaldol
followed by MPV reduction of the crotonaldehyde; Blue pathway shows the direct
MPV reduction of the resulting acetaldol....................................................................... 84
xiii
Figure 3.6 Free-energy profiles for the Prins condensation between ethylene and
acetaldehyde. Red pathway indicates a typical route of Prins condensation; Blue
pathway shows an additional proton diffusion step in between the reaction steps; Black
pathway shows the unlikely formation of MEK............................................................. 87
Figure 3.7 Free-energy profile for ethanol and acetaldehyde nucleophilic addition
reaction............................................................................................................................ 89
Chapter 4
Figure 4.1. Main reaction mechanism proposed for ethanol to 1,3-butadiene via
Toussaint’s aldol condensation ....................................................................................... 105
Figure 4.2. Conversion (●) and selectivity of main products (■ acetaldehyde; ▲
ethylene; ♦ 1,3-butadiene) at different WHSV. Reaction conditions: T=723 K, Qtot =
50 cm3/min, Mcat=0.2 g, P0EtOH = 2.72; 3.77; 5.15; 6.96 kPa........................................... 110
Figure 4.3. In-situ DRIFTS spectra acquired of dehydrated (temperature programmed
to 773 K at 10 oC/min under air and cooled down to 100 oC) MgO, MgO WK (1:1)
catalysts and SiO2. Only hydroxyl region of 3800 to 3200 cm-1 is shown. Spectra are
acquired at 100 °C............................................................................................................ 111
Figure 4.4. DRIFTS spectra of adsorbed (a) CO2 and (b) pyridine on WK (1:1) catalyst
at different temperatures to probe the catalyst’s basicity and acidity at relevant
temperatures..................................................................................................................... 114
Figure 4.5. In-situ DRIFTS spectra in the hydroxyl group region of 3800 – 3200 cm-1
acquired of ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol on WK (1:1)
catalyst. Sample vapor was adsorbed on the sample surface and temperature ramped
up from 373 to 723 K while spectra being recorded. In-situ DRIFTS dehydrated
catalyst spectrum at 100 °C was used as a reference....................................................... 118
Figure 4.6. In-situ DRIFTS spectra acquired of ethanol on WK (1:1) catalyst in 3200
to 1000 cm-1 spectral region. Ethanol was adsorbed on the sample surface and
temperature increased from 373 to 723 K while spectra being recorded. In-situ
DRIFTS spectra of the sample surface with no adsorbate present at every corresponding
temperature were used for reference. In-situ DRIFTS spectra acquired for ethanol
adsorbed on MgO are shown in the inset for 1200 to 1000 cm-1 spectral region............ 123
Figure 4.7. In-situ DRIFTS spectra acquired of acetaldehyde on WK (1:1) catalyst in
3200 to 1000 cm-1 spectral region. Acetaldehyde was adsorbed on the sample surface
and temperature increased from 373 to 723 K while spectra being recorded. In-situ
DRIFTS spectra of the sample surface with no adsorbate present at every corresponding
temperature were used for reference. In-situ DRIFTS spectra acquired for acetaldehyde
adsorbed on MgO are shown in the inset for 1200 to 1000 cm-1 spectral
region............................................................................................................................... 126
xiv
Figure 4.8. In-situ DRIFTS spectra acquired of crotonaldehyde on WK (1:1) catalyst
in 3200 to 1000 cm-1 spectral region. Crotonaldehyde was adsorbed on the sample
surface and temperature increased from 373 to 723 K while spectra being recorded. In-
situ DRIFTS spectra of the sample surface with no adsorbate present at every
corresponding temperature were used for reference........................................................ 127
Figure 4.9. In-situ DRIFTS spectra acquired of crotyl alcohol on WK (1:1) catalyst in
3200 to 1000 cm-1 spectral region. Crotyl alcohol was adsorbed on the sample surface
and temperature increased from 376 to 723 K while spectra being recorded. In-situ
DRIFTS spectra of the sample surface with no adsorbate present at every corresponding
temperature were used for reference................................................................................ 128
Figure 4.10. PBE optimized structures of ethanol (I), acetaldehyde (II), its enolate
conformation (II), crotonaldehyde (IV), crotyl alcohol (V) and 1,3-butadiene (VI) on
MgO surface low coordination Mg3cO4c or Mg3cO5c surface sites. Numbers refer to the
particular steps in catalytic transformation cycle shown in Figure 4.1........................... 134
Figure 4.11. In-situ DRIFTS spectra acquired of ethanol on WK (1:1) catalyst. Ethanol
was adsorbed on the sample surface, flown continuously and temperature increased
from 376 to 723 K while spectra being recorded. In-situ DRIFTS spectrum of the
sample surface with adsorbed ethanol present at 373 K was used for reference............. 139
Figure 4.12. In-situ DRIFTS spectra acquired of acetaldehyde on WK (1:1) catalyst.
Acetaldehyde was adsorbed on the sample surface, flown continuously and temperature
increased from 376 to 723 K while spectra being recorded. In-situ DRIFTS spectrum
of the sample surface with adsorbed acetaldehyde present at 373 K was used for
reference........................................................................................................................... 140
Figure 4.13. In-situ DRIFTS spectra acquired of crotonaldehyde on WK (1:1) catalyst
under ethanol vapor flow. Crotonaldehyde was adsorbed on the sample surface,
flushed with inert gas and ethanol was introduced under continuous flow with
temperature increased from 376 to 723 K while spectra being recorded. In-situ
DRIFTS spectrum of the sample surface with adsorbed crotonaldehyde at 373 K was
used for reference. For comparison, 523 K spectrum of crotonaldehyde adsorbed with
no gas phase present is shown in red dotted line............................................................. 143
Figure 4.14. Complete surface reaction scheme on ethanol reaction over MgO/SiO2
catalyst. (I) Crotonaldehyde, (II) adsorbed crotyl alcohol, (III) 1,3-butadiene, (IV) 2,4-
hexadienal, (V) paraldehyde, (VI) metaldehyde. ............................................................ 147
Figure S4.1. In-situ spectroscopy of ethanol on MgO catalyst. Ethanol was adsorbed
on the sample surface and temperature ramped up from 373 to 723 K while spectra
being recorded. Subtracted spectra are shown. Spectra are offset for clarity................ 148
Figure S4.2. In-situ spectroscopy of acetaldehyde on MgO catalyst. Acetaldehyde was
adsorbed on the sample surface and temperature ramped up from 373 to 723 K while
spectra are being recorded. Spectra are offset for clarity................................................ 148
xv
Chapter 5
Figure 5.1. Possible combination of metal atoms that act as Lewis acid sites: A: Mg3C
(open), B: Mg3C (closed), C: Mg4C (closed), D: Mg4C (open), E: Mg5C (open), F: Mg5C
(closed)............................................................................................................................. 159
Figure 5.2. Catalytic activity of MgSi-WK between 350-450°C. Inset: Arrhenius plot
of ethylene and 1,3-BD. Catalyst mass = 0.1 gr, total flow rate = 55 ml/min, pethanol =
2.5 kPa. ....................................................................................... ................................... 163
Figure 5.3. Comparison of XRD patterns of MgSi-WK and MgSi-IWI. WK with
different oxide ratios are overlaid for comparison. ......................................................... 164
Figure 5.4. Depth-profile of (a) MgSi-IWI and (b) MgSi-WK as probed using HS-
LEIS. HS-LEIS spectra of layer by layer sputtering of catalyst surface are shown in the
inset.................................................................................................................................. 165
Figure 5.5. Left: Comparison of OH groups of MgSi-WK and MgSi-IWI as probed by
in-situ dehydrated DRIFTS experiments; right: OH groups of WK catalysts with
different oxide ratios........................................................................................................ 167
Figure 5.6. TPRS spectra of ETB reaction over MgSi-WK with ethanol as the feed
(left) and acetaldehyde as the feed (right). EtOH: ethanol; AA: acetaldehyde; CA:
crotonaldehyde; C-OH: crotyl alcohol............................................................................. 169
Figure 5.7. TPRS spectra of ETB reaction over MgSi-WK with ethanol and
acetaldehyde as the coreactants. Acetaldehyde is pre-adsorbed on the surface and
temperature ramp is under ethanol flow.......................................................................... 171
Figure 5.8. Acid-base characterization of MgSi-IWI and MgSi-WK catalysts probed
using CO2 (left) and pyridine (right). Spectra at high temperature (450°C) and low
temperature (100°C) are shown....................................................................................... 173
Figure 5.9. In-situ acid-base characterization of MgSi-WK catalyst before and after
ethanol adsorption at 100°C and reaction at 200°C using CO2 (left) and pyridine
(right)............................................................................................................................... 174
Figure 5.10. Acid-base poisoning reactivity testing using (a) CO2, (b) propionic acid,
and (c) NH3 to determine the role of each site during ethanol conversion to 1,3-BD over
WK-800 MgO/SiO2 catalyst. Reactions are carried out at 425 °C, mcat = 0.1 g, pethanol =
2.5 kPa, total flow = 55 ml/min. All formation rates are normalized to initial 1,3-BD
formation rate................................................................................................................... 178
Figure 5.11. Productivity of (a) 1,3-BD, (b) ethylene, and (c) acetaldehyde of Na-
poisoned MgSi-WK catalysts between 350-450°C. Catalyst mass = 0.1 gr, total flow
rate = 55 ml/min, pethanol = 2.5 kPa................................................................................... 181
xvi
Figure 5.12. Bottom: DRIFTS characterization of Na-doped MgSi-WK using (a) CO2
and (b) NH3. Spectra are taken at 100°C after extensive evacuation with N2. Top: (a)
CO2 desorption spectra of 1000 ppm Na-doped MgSi-WK at 100, 300, and 450°C and
(b) NH3 desorption on 0 ppm and 250 ppm Na-doped MgSi-WK at 300°C. Spectral
subtraction was done using the spectra of the dehydrated catalysts at respective
temperatures as the background....................................................................................... 184
Figure 5.13. (a) MgO periodic model used for DFT simulation of NH3 adsorption on
MgO Lewis acid sites: (b) Mg3C, closed, (c) Mg4C, closed, (d) Mg3C, open, (e) Mg4C,
open. Multiple possible adsorption sites, i.e. kink (Mg3CO4C), edge (Mg4CO4C), and
planar (Mg5CO5C) are highlighted.................................................................................... 185
Figure 5.14. Schematic diagram to show the presence of various sites investigated with
NH3 and CO2 DRIFTS experiments. The basic sites (orange) are shown in the figure as
both Brønsted base (OH) and Lewis site (electron accepting oxygen atoms), and acid
sites (blue) are represented as Brønsted acid sites (H) and Lewis acid sites (electron
donating magnesium and silicon atoms).......................................................................... 186
Figure 5.15. Representation of the role of basic sites during ethanol conversion to
acetaldehyde. Top figure represents dehydrated (pretreated) catalyst; bottom figure
demonstrates the absence of bicarbonate when CO2 is adsorbed in-situ after reaction at
200 °C.............................................................................................................................. 188
Chapter 6
Figure 6.1. Local structure analysis of (a)MgO, (b)Cu-MgO, and (c)Zn-MgO. The
Bader atomic charge on each atom is indicated by the boldfaced numbers.................... 204
Figure 6.2. All stable intermediates and transition states calculated following the
reaction pathways. (1A-1C): ethanol dehydrogenation to acetaldehyde; (2A-2C):
ethanol dehydration to ethylene; (3A-3C): C-C bond formation step in acetaldehyde
aldol condensation to 3-hydroxybutanal (acetaldol); (4A-4C): C-C bond formation step
in Prins condensation of acetaldehyde and ethylene. Calculations are carried over
Zn/MgO model catalysts (prefix: Zn), and Cu/MgO model catalysts (prefix: Cu)......... 207
Figure 6.3. Potential energy surface for ethanol (a)dehydrogenation and
(b)dehydration over MgO, Zn/MgO, and Cu/MgO catalysts. (●)MgO, (■) Cu-MgO, (♦)
Zn-MgO........................................................................................................................... 212
Figure 6.4. Potential energy surface for first C-C bond formation via (a) acetaldehyde
aldol condensation and (b) Prins reaction between acetaldehyde and ethylene over
MgO, Zn/MgO, and Cu/MgO catalysts........................................................................... 213
Figure 6.5. Comparison of XRD patterns between CuMgSi, ZnMgSi, and MgSi......... 215
Figure 6.6. In-situ dehydrated DRIFTS of OH region of MgSi, CuMgSi, and ZnMgSi.
Spectra were taken at 100°C under He flow after pretreatment at 500°C for 1 hour.
Spectra were offset for clarity.......................................................................................... 216
xvii
Figure 6.7. In-situ UV-Vis DRS spectra of (a) dehydrated CuMgSi catalyst referenced
with Cu/MgO (CuMg), Cu/SiO2 (CuSi), CuO, and MgSi; (b) dehydrated ZnMgSi
catalyst referenced with Zn/MgO (ZnMg), Zn/SiO2 (ZnSi), ZnO, and MgSi. Inset: UV-
Vis spectra of different loadings of Zn on MgO/SiO2
catalysts........................................................................................................................... 217
Figure 6.8. Scanning Transmission Electron Microscopy images of ZnMg, ZnMgSi,
CuMg and CuMgSi samples. Energy Dispersive Spectroscopy profiles (smoothed) are
also provided. Small ZnO nanoparticles are shown in ZnMgSi with red
arrows.............................................................................................................................. 220
Figure 6.9. Productivity comparison of 1,3-BD (■), ethylene (●), and acetaldehyde
(▲) over (a) MgSi, (b) CuMgSi, and (c) ZnMgSi. Dotted lines are meant to guide the
eyes. Insets: Arrhenius plots to show apparent activation energies of the three
(by)products. Reactions are carried out between 325 - 450°C, mcat = 0.1 g, pethanol = 1.8
kPa, total flow = 55 ml/min............................................................................................. 224
Figure 6.10 Evolution of each peak during in-situ temperature-programmed ethanol
DRIFTS over (a) MgSi, (b) CuMgSi, (c) ZnMgSi. Insets: original spectra of ethanol
DRIFTS from where the peaks were deconvoluted......................................................... 227
Figure 6.11. In-situ UV-Vis DRS under constant ethanol flow over (a) CuMgSi and
(b) ZnMgSi...................................................................................................................... 231
Figure 6.12. Normalized XANES spectra of CuMg, CuSi, and CuMgSi (a) and Cu foil,
CuO, Cu2O, and CuMg (b). Inset: Cu K-edge k2-weighted EXAFS data of
corresponding spectra. XANES spectra in Figure 6.12(a) are offset vertically for
clarity............................................................................................................................... 232
Figure 6.13. Normalized temperature-programmed operando XANES spectra of
CuMgSi catalyst under He flow (top) and ethanol flow (bottom). Inset: enlarged region
of the pre-edge features to elucidate changes at different temperature........................... 235
Figure 6.14. Normalized time-resolved operando XANES spectra of CuMgSi catalyst
under ethanol flow at 400°C. Inset: enlarged region of the pre-edge features to elucidate
changes at different temperature...................................................................................... 236
Figure 6.15. Coordination number changes during reaction of ethanol to 1,3-BD over
CuMgSi............................................................................................................................ 237
Figure 6.16. XANES spectra of the simulated CuO Model 1: Cu in a local evironment
surrounded by 6 oxygen atoms and Model 2: Cu in a local environment surrounded by
4 oxygen atoms................................................................................................................ 239
Figure 6.17. (a) Normalized XANES spectra of ZnMg, ZnSi, ZnMgSi, Zn foil, and
ZnO. Inset: Zn K-edge k2-weighted EXAFS data of corresponding spectra. (b) Fourier
transforms of the EXAFS spectra of ZnMg, ZnO, and Zn
foil.................................................................................................................................... 242
xviii
Figure 6.18. Normalized temperature-programmed operando XANES spectra of
ZnMgSi catalyst under He flow (a) and ethanol flow (b). Inset: enlarged region of the
pre-edge features to elucidate changes at different temperature. (c) Temperature-
induced change in coordination number of Zn-Mg and Zn-O bonds during the
reaction............................................................................................................................ 243
Figure S6.1. XRD patterns of (a) Zn/MgO and (b) Zn/SiO2 at different
loadings............................................................................................................................ 247
Figure S6.2. XRD patterns of (a) Cu/MgO and (b) Cu/SiO2 at different
loadings............................................................................................................................ 247
Figure S6.3. In-situ DRIFTS of OH region of dehydrated MgSi catalysts references at
100°C for Cu-promoted (left) and Zn-promoted (right). Spectra are offset for
clarity............................................................................................................................... 248
Figure S6.4. Tauc plot of CuO (left) and deconvoluted Cu species of CuMgSi catalyst
(right) to determine the edge energy/band gap (E0) for correlation with number of Cu
coordination..................................................................................................................... 248
Figure S6.5. In-situ UV-Vis difference spectra of oxidative dehydration of (a) CuMgSi
and (b) ZnMgSi................................................................................................................ 248
Figure S6.6. Poisoning reactivity testing using CO2 to determine the role of basic sites
during ethanol conversion to 1,3-BD over (a) MgSi, (b) CuMgSi, and (c) ZnMgSi.
Reactions are carried out at 400 °C, mcat = 0.1 g, pethanol = 2.5 kPa, total flow = 55
ml/min. All formation rates are normalized to initial 1,3-BD formation
rate................................................................................................................................... 249
Figure S6.7. CO2 Temperature Programmed-DRIFTS on (a) MgSi, (b) CuMgSi, and
(c) ZnMgSi....................................................................................................................... 253
Figure S6.8. Poisoning reactivity testing using propionic acid to determine the role of
basic sites during ethanol conversion to 1,3-BD over (a) MgSi, (b) CuMgSi, and (c)
ZnMgSi. Reactions are carried out at 400 °C, mcat = 0.1 g, pethanol = 2.5 kPa, total flow
= 55 ml/min. All formation rates are normalized to initial 1,3-BD formation
rate....................................................................................................................................254
Figure S6.9. Poisoning reactivity testing using NH3 to determine the role of acid sites
during ethanol conversion to 1,3-BD over (a) MgSi, (b) CuMgSi, and (c) ZnMgSi.
Reactions are carried out at 400 °C, mcat = 0.1 g, pethanol = 2.5 kPa, total flow = 30 ml/min
(without NH3), 55 ml/min (with NH3). All formation rates are normalized to initial 1,3-
BD formation rate. NH3 desorption spectra on MgSi catalysts at 100°C are shown in
(d) .................................................................................................................................... 255
xix
Figure S6.10. DRIFTS spectra in the C-H stretching (left) and bending (right) region
of methanol desorption under He flow on unpromoted (top) and promoted (bottom)
catalysts............................................................................................................................ 257
Figure S6.11. Online MS analysis during operando methanol DRIFTS of CuMgSi,
ZnMgSi, MgSi and reference MgO................................................................................. 260
Figure S6.12. In-situ UV-Vis DRS of ethanol reaction on undoped MgO/SiO2 catalyst.
Difference spectra is shown, where catalyst spectra at 100°C with chemisorbed ethanol
is used as a reference........................................................................................................264
Figure S6.13. R-space EXAFS spectra of CuMg catalyst, in comparison to Cu foil,
CuO, and Cu2O................................................................................................................ 264
Figure S6.14. Corresponding MS data of in-situ XANES-EXAFS for ethanol to 1,3-
BD over (a) CuMgSi, (b) ZnMgSi................................................................................... 265
1
Abstract
Increasing concerns regarding global warming, which is caused by growing CO2
emissions, have led to efforts focused on discovering alternatives to petroleum for energy
and commodity chemical production. (Bio)ethanol has been seen as a platform molecule
with increasing production and versatility for upgrading to various high-value fuels and
chemicals. Among those high-value chemicals is 1,3-butadiene (1,3-BD), which has
demonstrated widespread applications in polymer synthesis and as an organic chemistry
intermediate. Its conventional methods of production rely on oil as a feedstock, hence
suggesting the need for alternative and more sustainable routes. Interest in the catalytic
conversion of ethanol to 1,3-BD, introduced in the 1940s by Lebedev, has been revived
and is now focused on the development of selective catalysts, thus minimizing the need for
the high cost separation between 1,3-BD and other (by)products, such as C2 and C4 olefins
and oxygenates. The main components of the catalyst for this system are MgO and SiO2,
where its reactivity and selectivity depend heavily on the method of preparation. This
system is still at an early stage of development, with a lot of disagreements on structure of
the catalyst, optimum ratio of Mg:Si, reactive intermediates, reaction mechanisms, and
kinetics.
Reaction mechanism was studied intensively using both theoretical (DFT) and
experimental (spectroscopy) methods. Initial screening of the reaction mechanism using
DFT with MgO defect site, i.e. kink, demonstrated that aldol condensation is more viable
thermodynamically than Prins condensation. In the reaction mechanism, dehydrogenation
of ethanol to acetaldehyde, an important reactive intermediate, is shown to be the rate-
determining step (RDS) of the reaction. Comparison of the potential energy barrier also
2
shows that acetaldol, the product of acetaldehyde self-aldolisation, dehydration competes
with its hydrogenation with an ethanol molecule. This mechanistic study is also supported
by comprehensive in-situ DRIFTS. MgO/SiO2 catalyst is synthesized using a wet-kneading
method, with equivalent oxide mass ratio and thoroughly characterized with HS-LEIS,
DRIFTS, and XRD. Chemical probing was also done with different probe molecules, such
as pyridine, NH3, CO2, and methanol. Combination of several reactants and intermediate
shows that acetaldehyde is spontaneously transformed to crotonaldehyde under constant
reactant flow, while in-situ ethanol DRIFTS requires contribution from the gas-phase
ethanol to make 1,3-BD. Furthermore, the crotonaldehyde does not transform to 1,3-BD
under inert flow, it requires the presence of ethanol to complete the transformation to 1,3-
BD.
The resulting catalyst was extensively probed and characterized, revealing a silica-
rich surface, where comparison with incipient wetness impregnation catalyst shows a rather
Mg-rich surface. Surface silicate that is formed is confirmed by in-situ DRIFTS, where
new OH groups were formed. The basicity of the catalyst also varies significantly with
different methods of preparation and calcination temperature. All strong, medium, and
weak basic sites were found on the catalysts surface. More superior performance, however,
is shown to be enforced by lower amount of strong basic sites. Ammonia probing reveals
the presence of both open and closed Lewis acid sites (LAS) and limited amount of
Brønsted acid sites (BAS). Pyridine, on the other hand, could not identify any BAS, which
is due to its larger molecule size. This further demonstrates that the LAS on the catalyst is
much more accessible than the BAS.
3
Promotion of the catalyst with transition metal was shown to have a significant
enhancement on the reactivity. Since the RDS was determined to be the dehydrogenation
of ethanol, transition metal sites lower this barrier, and shift the RDS. Zn and Cu, two very
promising ethanol dehydrogenation catalysts, were separately impregnated on the
uncalcined wet-kneaded MgO/SiO2 support at low loadings, 2.5 and 1%, respectively. The
catalysts were thoroughly characterized using in-situ UV-Vis, methanol operando
DRIFTS, in-situ XANES and EXAFS, TEM, TPSR, and in-situ DRIFTS. Cu(II) exists as
a surface species coordinated in a tetrahedral geometry, where it has 0.8 (or ~1) nearest
neighbor, i.e. number of Cu-O-Cu bonds. The transition metal also possesses Cu-O-Mg
bond, hinting to formation of solid solution. Similar interaction was also observed for Zn,
suggesting the stronger interaction with Mg, instead of Si. This structural change affects
the basicity and acidity of the catalyst. Both CO2 and methanol probing with DRIFTS show
that the promoted catalysts have less affinity with CO2, while the BAS was eliminated,
replaced with another distinct LAS. Redox capability was also modified, shown by the
enhanced strength of the redox site in expense of its reduced quantity. During the reaction,
Cu(II) is reduced to Cu(0) via an intermediate Cu species, before the catalyst deactivates
after long hours of experiment. Zn, on the other hand, maintained its structure even after
extensively tested.
4
Chapter 1
Introduction
1. Background .......................................................................................................... 4
1.1. ETB Reaction Network ............................................................................. 11
1.1.1. Reaction Intermediates and Byproducts ................................................... 11
1.1.2. Proposed Reaction Mechanisms ............................................................... 12
1.2. Catalytic Systems ..................................................................................... 15
1.2.1. Reaction Conditions and Catalytic Performance ...................................... 16
1.2.2. MgO/SiO2 Catalysts ................................................................................. 18
1.2.3. ZrOx –based Catalysts ............................................................................... 24
1.2.4. Other Catalysts ......................................................................................... 26
2. Approach ............................................................................................................ 29
2.1. Approach .................................................................................................. 29
2.2. DFT Calculation ....................................................................................... 31
2.3. In-situ and Operando Spectroscopy ......................................................... 32
2.3.1. Infrared Spectroscopy ............................................................................... 33
2.3.2. UV-Vis Spectroscopy ............................................................................... 34
2.3.3. Operando XANES and EXAFS ................................................................ 35
2.3.4. Temperature-Programmed Reaction Spectroscopy .................................. 36
2.4. HS-LEIS ................................................................................................... 37
2.5. Probe Molecules ....................................................................................... 38
2.6. Product Determination with GC-MS ........................................................ 39 3. Thesis Outline ..................................................................................................... 41
References ....................................................................................................................... 43
1. Background
The growing environmental concerns caused by the increasing CO2 emissions have
incentivized endeavors on discovering alternatives for energy and chemical production.
While potential alternatives such as wind, solar, and nuclear had been able to partially
replace the need for power generation, biomass remains one the only options to mitigate
the petroleum consumption in chemical production section.1 Biomass valorization had
5
been extensively studied, with focuses on lignin depolymerization, sugar modification, and
hemicellulose fermentation to ethanol.2,3 The target molecules varied from aromatics,
alcohols, to new molecules that were envisioned to replace the incumbent from petroleum.2
While the fight for biomass upgrading is still far from over, (bio)ethanol has presented a
very interesting alternative, due to its abundance3 and its versatility to be upgraded to
different other higher-valued platform molecules or commodity chemicals. Figure 1.1
shows the ever-increasing production of ethanol and its upgradeability to different
molecules.
Ethanol upgrading to higher-valued chemicals has recently been pursued.4,5 It is
fairly reactive, and the fact that it has two carbons makes it relatively selective to even-
numbered carbon containing molecules. Different target were investigated, such as
hydrogen,5–7 n-butanol,8–13 ethylene and diethyl ether (DEE),14–20 propylene,21–25
isobutene,26–28 ethylene oxide,29,30 acetaldehyde,31–35 ethyl acetate,36–42 and 1,3-butadiene
(1,3-BD). Steam reforming had also been explored to replace the current energy-intensive
methane steam reforming processes. The ethanol to chemical processes is different from
Figure 1.1. Ethanol production rate increase from 2010 to 2017 (adapted from U.S.
Energy Information and Administration) (left) and ethanol upgrading map to different
highly valued chemicals (right).
6
methanol to olefin, where the latter is initiated by carbon pooling mechanism, activating
the surface methoxide and breaking the stable C-H bond. Most of the proposed mechanisms
for ethanol upgrading involve acetaldehyde formation, which can be activated by enolate
formation. This reaction itself opens a new pathway for the C-C bond formation, which is
also extensively studied.43
Hydrogen is a very clean fuel energy source, where its application will give off
water as the only product.44 Current hydrogen production is dominated by methane steam
reforming over Ni catalyst, which is also followed by water-gas shift reaction.45 However,
such reaction generally applies severe condition, 700-1100 °C, leading to the very high
capital costs for its facilities.45 Another pathway to make hydrogen is the photocatalytic
process, where water is split into hydrogen and oxygen. However, charge recombination
and thermodynamics stand as a major obstacle in achieving respectable yield.46 Ethanol
steam reforming, on the other hand, possesses a major advantage in the much lower
reaction temperature, <600 °C, some of the processes even had reaction temperatures of
250 °C.5–7 Partial oxidation of ethanol, i.e. autothermal reforming, is also another
alternative that is being investigated, due to it being much less energy intensive than the
endothermic steam reforming. Major challenge in this reaction remains the expected
carbonaceous deposit on the catalyst surface, which will lead to catalyst deactivation. This
carbonaceous deposit, however, can be mitigated to a certain extent by using suitable
supports, such as MgO, ZnO, CeO2, and La2O3.6
Other endeavors had been focused on valorizing ethanol into ethylene, propylene,
ethylene oxide, 1,3-BD, and n-butanol, which are among the top 30 industrial organic
chemicals based on weight produced in the USA.1 Ethylene is typically produced from oil
7
cracking, which also produces 1,3-BD as the byproduct. Popular alternative of this process
has been the dehydrogenation of ethane, due to the increasing production of ethane from
both shale gas and the byproduct of the paraffin from the oil cracking.4 Ethanol dehydration
was touted as a possible route for the production of bioethylene, where it has recently been
made economically possible due to the recent advances in the heterogenous catalysts, as
well as the lowered ethanol price.4 The dehydration of ethanol is carried out over solid acid
catalysts, such as SAPO-34,17 γ-Al2O3,14–17 and zeolites.14,17,18 γ-Al2O3 is the most stable
catalyst, whereas other catalysts were found to give higher ethylene yield and operate at
lower temperature. Promotion with transition metals, such as Ni17 and Mo,18 were found to
have prolonged the catalysts’ life, where the latter was shown to be more of a sacrificial
transition metal oxide to be reduced during the reaction.18 Lewis acid sites (LAS) were
observed to be the main site for ethanol conversion, while ethylene production was
maximized by the presence of medium and weak acid sites.16,17 Water, byproduct of the
ethanol dehydration, however, plays an important role in deactivating the catalyst, since it
was found to block the neighboring site, which prevents the C-H bond breaking to make
the C=C bond.19 The major challenge remains in limiting the bimolecular dehydration route
to make diethyl ether, which is much more thermodynamically favored than ethanol. This
can be done by using catalyst with confinement effects.20
Another C2 molecule that can be directly synthesized from ethanol is acetaldehyde.
This molecule had traditionally been used as an intermediate that is further converted to
other chemicals, mainly to acetic acid. However, this process was largely abandoned due
to the more selective Monsato and Cativa processes from methanol.47 Production of
acetaldehyde from ethanol follows two routes, partial oxidation and non-oxidative
8
dehydrogenation. The latter is becoming more attractive due to the production of hydrogen
as its byproduct, even though the presence of oxygen significantly enhances the activity.
Non-oxidative dehydrogenation of the catalyst was studied with different Cu catalysts,
including Cu/SiO2,31 Cu-Cr2O4,
32,33 and CuO/RHA (rice husk ash).34,35 The presence of
chromium stabilized the catalyst by preventing sintering during both reduction and
reaction.32,33 Significant improvement of acetaldehyde yield was achieved when Cu was
supported on rice husk ash, a silica-rich byproduct of domestic agriculture, resulting in
lowered Cu particle size and high surface area catalyst.34,35
Related to acetaldehyde, ethyl acetate is another product of ethanol upgrading that
involves the acetaldehyde as the reactive intermediate. Important catalysts that were
developed are Cu-Zn-Zr-Al-O, 36–38 supported Pd catalysts,39 Au/TiO2,40 and supported Cu
catalysts.41 The presence of different metals were reported to have different effects on the
catalyst.42 Zr and Al, for instance, enhanced the conversion of ethanol, with Zr favored
ethyl acetate, while Al favored dehydration products such as DEE and methyl ethyl ketone
(MEK). Collaborative effects of Zn and Zr were touted to limit the MEK, which is an
unwanted byproduct, while Cu increased the dehydrogenation reaction, resulting in high
ethyl acetate yield.42 There is still a lot of room for improvement for this system, since the
reaction mechanism is yet to be proven. Rational design of the catalysts is still far from
reach, shown by the previous investigators’ attempts to use various different transition
metals in their catalysts.
C4 molecules that are upgraded by creating new C-C bond, such as isobutene, n-
butanol, and 1,3-BD, are very attractive due to its higher value and its multiple applications.
Isobutene is an important commodity chemical, in particular as an additive to jet fuel and
9
as a raw material for various polymers. Pathway from ethanol was recognized by
converting it into acetone, and further converting acetone into isobutene. The highlighted
two-step reaction, however, requires basic and acidic catalysts. Bifunctional catalysts
containing balanced amount of base/acid sites were synthesized at Pacific Northwest
National Laboratory (PNNL).26–28 The catalysts, possessing two different sites, were able
to catalyze the two reactions in one pot, with basic and redox sites catalyzing the ethanol
to acetone, while the acid sites condensing the acetone into isobutene. Detailed study of
the catalysts revealed the importance and detrimental effects of Brønsted acid sites (BAS),
which catalyzed the second reaction, as well as isomerizing and polymerizing isobutene
into other butenes and coke.28
N-butanol has a lot of advantages over ethanol as a drop-in fuel. It has higher
solubility in gasoline (longer chain) and significantly higher calorific value (29.2 vs 19.6
MJ/dm3). Upgrading ethanol to n-butanol had been a focus of several research groups,8–13
and had converged into two classes of catalysts: hydroxyapatite and hydrotalcite. N-
Butanol has been traditionally produced from aldol condensation of acetaldehyde followed
by the catalytic hydrogenation and from oxo process.47 Revisiting the process is of
paramount interest due to the lowered ethanol price. Just like isobutene, the catalyst needs
to be bifunctional in nature. Very early attempts used combination of strong basic and
acids, such as alkali cation-exchanged zeolites48 and Na-promoted zirconia.49 Magnesia-
based catalysts present a unique class of bifunctional catalysts, in which they do not
necessarily possess strong basic sites, but still provide weak acid sites. Magnesia alone was
observed to give respectable n-butanol yield,8,10 while combination with alumina gave a
hydrotalcite structure, which increased the activity significantly.50–52 More improvement
10
was achieved by hydroxyapatite catalysts, calcium phosphate material, which was done by
tuning the Ca/P ratio, as stronger basic sites were shown to increase the n-butanol yield.53
Latest works had focused on the study of the material, where kinetic models and active
sites were determined using combination of several methods, such as SSITKA and in-situ
titration with probe molecules.12,13
Among all of the higher-valued target molecules, 1,3-BD represents the most
interesting target molecule. Back in the 1940s, both the US and the USSR produced 1,3-
BD from ethanol via two-step54,55 and one-step processes,56 respectively. In the two-step
process, ethanol was first converted into acetaldehyde before the product and a separate
ethanol feed were flowed to a second reactor containing a supported tantalum oxide
catalyst.54,55 The one-step process, however, incorporated all reactions in one-pot, using a
catalyst that was revealed to have both magnesia and silica.56 The process was abandoned
when catalytic cracking of oil became popular, since 1,3-BD were produced efficiently as
a byproduct of ethylene production. However, in the recent year, the shift of raw material
for ethylene production from oil to shale gas had decreased the availability of 1,3-BD,
prompting the revisiting of this old process. Improving the catalyst’s activity and stability
had been the main focus of the research, where new systems had also been established,
such as promoted MgO/SiO2,57–60 and solid Lewis acid catalyst systems.61,62
This chapter provides a review of ethanol to 1,3-BD (ETB) reaction system. This
review includes proposed reaction mechanisms, current state-of-the-art of the catalysts, and
the proposed active sites. Following the review, this chapter will discuss the techniques
that were used to study the system, as well as an outline of the thesis.
11
1.1. ETB Reaction Network
1.1.1. Reaction Intermediates and Byproducts
As shown in Figure 1.1, ethanol can be converted to a variety of other chemicals.
Optimizing ethanol to 1,3-BD requires analysis of thermodynamics to select the reaction
temperature and contact time, i.e. space velocity. Byproducts that are most commonly
found during the reactions are ethylene, CO2 and acetaldehyde.63 Other byproducts that
were occasionally observed are diethyl ether, butenes, propylene, C2-C4 paraffins, diols,
acetone higher alcohols and heavy oxygenates, such as n-butanol, n-hexanol,
crotonaldehyde, and hexadienal.58,64 Dictation by thermodynamics also depends on the
catalysts employed. At different reaction temperature, acid catalysts, such as γ-Al2O3,65
will give different product distributions from transition metal catalysts, such as
CuO/SiO2.35
Optimizing the performance of the catalyst also means suppressing the formation
of unwanted byproducts, such as butenes, ethylene, and acetaldehyde. Butenes are
particularly unwanted, since at high concentration, they will form azeotrope with 1,3-BD,
which makes separation costly.60 Ethylene, on the other hand, is of much lower-value, and
its formation is followed by the production of water, which was found to poison the catalyst
by competitive adsorption.19 Another byproduct of this system is acetaldehyde, shown to
be a reactive intermediate during the reaction, which will be explained in subsequent
subchapters. Thermodynamics also determines the selectivity between acetaldehyde and
ethylene, since ethanol dehydrogenation was found to be much more endergonic reaction
than ethanol dehydration.66 Furthermore, the reactivity of acetaldehyde to undergo further
12
reactions, such as polymerization, esterification, and aldol condensation, requires a careful
design of the catalyst to minimize the unwanted reactions.63,67
1.1.2. Proposed Reaction Mechanisms
Various byproducts produced during the reaction led to different reaction
mechanisms being proposed. In general, there had been five proposed reaction
mechanisms. The first two mechanisms, proposed by Lebedev4,56 and Ostromislensky68,69
were ruled out due to the improbable steps in their mechanisms. Briefly, Lebedev proposed
a reaction mechanism involving free radicals in a very complex sequence, while
Ostromislensky’s mechanism entailed reaction between ethanol and acetaldehyde, which
is followed by rearrangement of the hemiacetal to diols. Equation 1.1 shows the
stoichiometry of the reaction:
2 C2H5OH → C4H6 + 2 H2O + H2 (1.1)
Subsequently, Toussaint, et al. proposed reaction mechanism which involved
dehydrogenation of ethanol to acetaldehyde, followed by aldol condensation of two
acetaldehyde molecules, and dehydration of the aldol to give crotonaldehyde, which would
further undergo Meerwein-Ponndorf-Verley (MPV) reduction with ethanol and
dehydration to give 1,3-BD.55 Other reaction mechanisms that had been proposed were
called Prins condensation, proposed by Gruver, et al.,70 and carbanion mechanism, very
recently proposed by Chieregato, et al.71 Figure 1.2 comprehends the proposed reaction
mechanisms by Toussaint, et al., Gruver, et al. and Chieregato, et al.
13
The reaction mechanism proposed by Toussaint had been touted as the generally
accepted mechanism.4,5,72,73 Aldol condensation and the subsequent dehydration steps were
reported to be facile, while the rate of ethanol dehydrogenation and MPV reduction step of
crotonaldehyde depends on the catalyst. Over MgO/SiO2 catalysts, for instance, the rate-
determining step was ethanol dehydrogenation,66 while for Lewis acid catalysts, MPV
reduction of crotonaldehyde with ethanol as the hydrogen source was thought to be the
rate-limiting step.72 There are still, however, several issues with the mechanism, including
Figure 1.2. Proposed reaction mechanisms for ethanol conversion to 1,3-BD: (a)
Toussaint’s aldol condensation; (b) Gruver’s Prins condensation; (c) Cavani’s carbanion
mechanism.
(a)
(b)
(c)
14
the very rare occurrences of the reactive intermediates mentioned.71 Furthermore, acetaldol
was also shown to decompose to two acetaldehyde molecules when introduced to Ta/SiO2
catalyst.55 The very rare occurrence of the reactive intermediates was due to the rapid
reaction steps at the mostly employed reaction temperature. Second, acetaldol is a very
unstable molecule, it could either dehydrate to crotonaldehyde and water when heated, and
therefore, data analysis of the reaction should be carried out with extra caution.
The other rejected mechanism is the Prins mechanism, where ethanol undergoes
two parallel reactions, i.e. dehydrogenation to acetaldehyde and dehydration to ethylene.70
The problem with this mechanism is the ethylene protonation, an indispensable step during
the Prins mechanism, that leads to a highly unstable carbocation,4 and when this step was
not involved, the coadsorption of both ethylene and acetaldehyde on the surface is very
unstable.66 An alternative mechanism was proposed by Chieregato, et al., where ethanol
dehydrogenation resulted in two different entities.71 If the step was preceded by
dissociative adsorption, ethanol would be converted to acetaldehyde. Otherwise, a
physisorbed ethanol molecule would break a C-H bond and be converted to a stabilized
carbanion.71 This carbanion would further react with acetaldehyde to make 1,3-BD or with
ethanol to make n-butanol and 1-butene. Several issues were readily identified with this
mechanism. First issue was the use of bare MgO as the catalyst, which, if not
hydrothermally treated, would not give off 1,3-BD.71 Second, our attempts of using bare
MgO as the catalyst for Diffuse Reflectance Infrared Fourier Transformed Spectroscopy
(DRIFTS) study did not show any formation of the surface intermediates that were
demonstrated by their reports.63 Bare MgO exhibits very high absorption of CO2, which
would hamper identification of the surface intermediates that are located at similar
15
wavenumber ranges. Third, the experiment was carried out under inert flow, and due to the
nature of the cascade reaction, the amount of ethanol would be insufficient to convert
further into 1,3-BD, given all the ethanol will be consumed to acetaldehyde. Furthermore,
the peak fittings were done with solely focusing on the intermediates, without considering
the possibility of the side reactions of acetaldehyde, such as acetate formation, aldol
condensation, and polymerization. Finally, the model used in the density functional study
(DFT) study did not necessarily represent the true condition of the surface. The DFT
calculation was done on a cluster MgO, with a small number of atoms used to represent
the catalyst. Stability calculation was not done either in choosing the right defect site, with
corner MgO being the model site. Later attempt by our group demonstrated that the
postulated carbanion formation does not take place when C-H bond is broken, instead, the
resulting TS further broke the C-O bond to give-off ethylene, which disputed their
proposed mechanism.66
1.2. Catalytic Systems
There had been a lot of efforts to improve the catalytic performance of the catalysts
for the synthesis of 1,3-BD from ethanol.72,73 As mentioned previously, apart from the use
of MgO/SiO2-based catalysts,57,74–77 efforts had been carried out for other classes of
catalysts, such as zirconia-based catalysts,61,78 and mixed metal oxide catalysts.79–83 The
complex reaction mechanism calls for a very demanding catalyst specification; the
catalysts have to possess balanced amount of basic, acidic, and redox sites. Typically, there
are several ways to improve the catalysts’ performance, by promotion with other metal
oxides, hydrothermal or chemical post-synthesis treatment, and modification of the
16
preparation methods. This section will thoroughly discuss the current state-of-the-art of the
catalytic system.
1.2.1. Reaction Conditions and Catalytic Performance
1,3-BD had been the most realistic target molecule for ethanol upgrading, mainly
due to the abundance in ethanol supply and decline in 1,3-BD supply.60,84 Once a
competitive process, the catalytic cracking of petroleum feed had taken over the production
method, which in turn, demanded the green chemistry process to be a lot more efficient.
For the process to be competitive, there are several factors that have to be considered.
Among of these factors, 1,3-BD productivity and catalyst stability are the most important
factors. The minimum requirement for 1,3-BD productivity was suggested to be 0.15 gBD
gcat-1 h-1,83 while the catalyst needs to be stable for long hours of production, since catalyst
regeneration can also be very costly. Deactivation typically was due to coke deposition on
the catalyst, coupled with poisoning from water formed by the dehydration reaction.
Another factor that also plays important role is the reaction temperature, where
typically, the reaction takes place at 300 and 400 °C for low weight-hourly space velocity
(WHSV) of 0.2-1 h-1. The optimum reaction condition itself varied due to the nature of
different catalysts employed. For instance, unpromoted MgO/SiO2 catalysts typically
require higher operating temperature than promoted MgO/SiO2 catalysts, due to the higher
activation temperature for ethanol dehydrogenation to acetaldehyde.58,59 Other relevant
conditions are pressure and WHSV, which is essentially a unit to define catalyst to reactant
ratio. The reaction had been carried out under atmospheric pressure in a fixed bed reactor,
with the exception from Bhattacharyya, et al., who have used fluidized-bed reactor.85
Additional consideration is the ethanol to acetaldehyde ratio in reactant feed. This
17
parameter is also important for one-step process, since it will be favorable to recycle some
of the acetaldehyde back to the reactor. Table 1.1 shows the catalytic performances of
recently investigated catalysts.
No Catalyst T (°C) WHSV
(h-1)
TOS
(h) X (%)
YBD
(%)
PBD (gBD
gcat-1 h-1)
Ref.
MgO/SiO2 catalysts
1 MgO hydrothermally-
treated 400 0.2 7.0 36.6 17.2 0.02 86
2 WK MgO-SiO2 (1:1) 425 1.1 4.0 52.0 17.7 0.23 58
3 MC MgO-SiO2 (1:1) 400 1.2 3.3 10.0 3.6 0.02 59
4 SG MgO-SiO2 (4:1) 400 0.4 sa N/A 53.0 14.6 N/A 87
5 WK MgO-SiO2 (1:1) 400 0.1 3.0 65.9 32.5 0.19 88
6 MC MgO-SiO2 (1:1) 400 1.0 N/A 41.2 23.6 0.14 89
7 IWI MgO-SiO2 (1:1) 300 1.1 3.3 ~3.5 ~1 0.01 60
8 WK MgO-SiO2 (13:7) 450 4.1 1.0 95.0 73.2 1.15 77
9 1% CuO/MgO-SiO2 425 1.1 4.0 74.0 36.3 0.48 58
10 2% Ag/MgO-SiO2 400 1.2 3.3 50.0 20.5 0.15 59
11 5%Ga/MgO-SiO2 (1:1) 400 0.1 3.0 98.8 52.4 0.31 88
12 2% Zn/MgO-SiO2 425 1.0 3.0 84.6 45.0 0.26 57
13 0.1% Na/MgO-SiO2 350 0.2 N/A 100.0 87.0 0.08 90
14 1.2% Zn/Talc 400 8.4 7.0 41.6 21.3 1.06 86
15 3% Au/MgO-SiO2 (1:1) 300 1.1 3.3 45.0 27.0 0.14 60
16 1.5% Zr 1% Zn/MgO-
SiO2 (1:1) 375 0.6 3.0 40.0 30.4 0.13 91
17 1.2% K/ZrZn/MgO-SiO2
(1:1) 375 0.6 3.0 26.0 27.1 0.12 91
Zr/SiO2 catalysts
1 2% Ag/4% Zr/SiO2 320 0.31 5.0 55.2 39.4 0.07 62
2 1% Ag/Zr/BEA
(Si/Zr=263) 320 0.64 3.0 30.1 19.2 0.07 61
3 3.5% Ag/Zr/BEA 320 1.2-3 3.0 - - 0.59 92
4 2000 ppm Na Zn1Zr10On 350 6.2 30.0 54.4 14.1 0.49 93
5 2% ZnO/7% La2O3/1%
ZrO2/SiO2 400 2 N/A 100.0 60.2 0.71 94
Other catalysts
1 3% Hf/9.3% Zn/HM 360 0.64 10.0 98.6 68.4 0.26 79
2 1% Cu/1% Ta/SiBEA
(Si/Al=1300) 325 0.5 3.5 87.9 63.9 0.19 95
Table 1.1 Catalytic performance of different catalysts studied for one-step ethanol to 1,3-
BD conversion. a Contact (residence) time; WHSV (weighted-hourly space velocity);
TOS (time-on-stream); X (ethanol conversion), Y (yield); P (productivity)
18
1.2.2. MgO/SiO2 Catalysts
The original catalyst employed by Lebedev in his one-step process was disclosed
by Natta and Rigamonti to contain MgO and SiO2.96 MgO by itself is a very strong basic
catalyst,97,98 and its interaction with SiO2 were suggested to have formed distinct LAS,
making it a bifunctional catalyst.99 Furthermore, the defect sites of MgO are essentially a
very strong electron acceptor, making it a very strong Lewis acid-base pair.66 Proper
amount of acid-base sites was determined by changing the MgO to SiO2 ratio75,77,87,99 and
by different preparation methods.60,99,100 A contradicting result, however, was recently
reported by Baba, et al., where a bare MgO that was hydrothermally treated resulted in a
conversion of 36% and 1,3-BD selectivity of 47%.86 This MgO was reported to have a very
distinct showed that SiO2 presence was not indispensable to the catalyst.
The main issue with the catalyst system is that it is highly influenced by the Mg to
Si ratio and preparation methods. Discrepancies can be found on every considered
parameters, including the Mg to Si ratio and which preparation methods that yielded the
best performing catalysts. For instance, wet-kneading (WK),58,99 incipient-wetness
impregnation (IWI),60 mechanical mixing,59,89 sol-gel,87 and hydrothermal synthesis,86
were each reported as the best preparation method. Reports for optimum Mg to Si ratio
also contained a lot of disagreements, even for the same preparation method. For sol-gel
catalyst this parameter was reported to be between 9 to 15,87 both 1:189 and 2:159 for dry
milling, and 1:158,99 (17:3)75 for wet-kneading. In the case of promotion with transition
metals, the order of promoters impregnation and calcination also altered the activity.57,58,60
Works on unpromoted MgO/SiO2 catalysts mainly concerned the method of
preparation and oxide ratio. Wet-kneading method had been reported as the best
19
preparation method.58,77,99 Wet-kneading was defined as “a process in which two or more
solid precursor materials are combined and stirred (mechanically or magnetically)
thoroughly in a liquid medium.”76,101 This method was first used by Natta in 1947,101 and
maintained its relevance through the work of Niiyama, et al.,75 Ohnishi, et al.,90 Kvisle, et
al.,99 and Angelici, et al.76 The wet-kneading was typically done with either MgO102 or
Mg(OH)258,90 as the starting material, and typically in water, since other solvent, such as
ethanol, had exhibited unwanted effect.101
Study by Weckhuysen’s group showed that wet-kneading method is superior to co-
precipitation and mechanical mixing.58 Wet-kneading method resulted in 1,3-BD yield of
~17%, significantly higher than co-precipitated catalysts (~8%) and mechanically mixed
catalyst (<1%), with 24 h time-on-stream (TOS). Co-precipitated catalyst produced
significantly higher amount of ethylene, attributed to its higher ethylene selectivity. On the
other hand, wet-kneading resulted in a layered magnesium silicate phase, which was
correlated to the higher activity of the catalyst.76 A detailed acid-base characterization
using Hammett indicator, DRIFTS with probe molecules, and TPD further produced
certain criteria for a good catalyst.100 In particular, strong basic sites have to be limited,
with participation of mostly medium and weak basic sites, in combination with some Lewis
acidity. Co-precipitated catalyst possessed combination of both strong acid-base sites,
which are shown to have increased the generation of ethylene.100
Mechanical mixing (dry milling) with different silica materials were studied by
Jannsens, et al.59 Two ordered mesoporous materials (COK-12 and MCM-41) were used
along with amorphous mesoporous SiO2. MgO was first hydrated to form Mg(OH)2, which
was followed by dry milling with the SiO2 material. The resulting dry mixture was then
20
wetted with water before being dried and calcined. The wetting process appeared to have
a significant effect on the acidity-basicity of the catalyst. This process dispersed MgO,
further creating additional Lewis acid-base sites.59 Characterization of the catalyst showed
silica support covered by magnesia flakes in the case of SiO2. The other silica supports, i.e.
COK-12 and MCM-41, showed a loss of mesoporous structure when mixed with magnesia
with this preparation method. These catalysts did not show significant activity toward 1,3-
BD, with the MgO/COK-12 and MgO/MCM-41 being the worst catalyst due to their
collapsed structure. The diffusion of MgO into the lattice was suggested to limit access of
reactant to the active sites, which was also aggravated by the presence of large magnesium
silicate phases.59
Sol-gel was another explored preparation method.87 Starting from Mg/Si oxide ratio
of 2 up to 9, significant fosterite, a magnesium silicate, phase was identified using X-ray
diffraction (XRD) and attenuated total reflectance infrared spectroscopy (ATR-IR). This
fosterite phase, however, was detrimental to the process, since it possessed more acid sites,
leading to an enhanced ethylene yield. High amount of Mg, i.e. Mg/Si > 15 accumulated
the alkenol intermediate, due to the limited amount of acid sites. The best performing
catalysts exhibited combination of a limited number medium-strength acid sites with strong
basic properties, which contradicted Weckhuysen’s finding.87,100 The LAS, however, was
observed to be transformed into BAS when water was formed as the byproduct.
MgO/SiO2 with various MgO loading was synthesized using IWI method.60 For
loading between 10-80%, a magnesium silicate hydrate (MSH) phase was observed at
~50% loading. Higher loading at 80% showed a crystalline MgO phase, presumably
formed by excessive MgO formation covering SiO2. 29Si magic-angle spinning (MAS)
21
NMR spectroscopy showed that at lower loading at 10 and 30% MgO, SiO2 still retaine its
support characteristics, where Mg2+ cations mainly interact with the isolated silanol groups.
The Q3 feature was caused by silanol groups [Si*(OSi)3(OH)] and [Si*(OSi)4]. However,
at higher loading, new peaks associated with [Si*(OMg)(OSi)3], [Si*(OMg)(OSi)2(OH)],
and [Si* (OMg)2(OSi)2] were formed. As expected, higher MgO loading increased the
basicity of the catalyst, demonstrated by CO2 DRIFTS. The reactivity study showed that
higher loading significantly reduced the catalyst’s conversion while increasing selectivity
toward both acetaldehyde and 1,3-BD. The high silanol content for lower loading was
suggested to be responsible for ethanol dehydration at lower loading.
Other unpromoted MgO/SiO2 systems that have been studied are clay materials and
talc.86,103,104 Without promotion with transition metal oxides, clay, a naturally occurring
magnesium silicate mineral, favored dehydration of ethanol to produce ethylene.103,104 Talc
is a 1:2 layered structure, where a unit cell includes six octahedral sites and eight tetrahedral
sites. Mg2+ ions represent the former, while cations for the latter are Si4+. Significant
improvement of these magnesium silicate materials was attained by promotion with
different transition metal catalysts. Nickel,104 manganese,103 and zinc86,105 were all
incorporated into the catalyst to achieve higher conversion. Characterization of Zn/talc
carried out using ICP, XRD, and XPS suggested that Zn was incorporated into the lattice.
The Zn site in the catalyst was shown to be responsible for the enhanced ethanol
dehydrogenation step, which increases the overall 1,3-BD yield.86
Promotion using other metal oxides were very commonly done to improve 1,3-BD
activity. Mainly, this was done to add redox sites to the catalysts and hence, to reduce the
energetic barrier of the rate-limiting step. Larina, et al. published a study on the role of
22
ZnO as a promoter for magnesia/silica catalyst on the ethanol conversion to 1,3-BD.57 The
amount of both Lewis and BAS were investigated using pyridine adsorption spectroscopy.
The experiment revealed that there are two types of LAS in the catalyst, one in MgO-SiO2
contact phase and one in ZnO-SiO2. When tested over MgO single oxide, the formation of
1-butanol was preferred, indicating the necessity of both acidic and basic sites in the
catalysts. However, the excessive silica content would lead to the preferential formation of
ethylene, since the number of acid sites would significantly be increased. Catalytic testing
conducted demonstrated the enhanced yield of 1,3-BD attributed to the ZnO role in
catalyzing dehydrogenation of ethanol as the first step of the reaction. The author
postulated this step to be the rate-determining step for the complete mechanism 57.
Other transition metal oxides have also been used in the past to improve the
MgO/SiO2 catalysts’ performance. Angelici thoroughly studied 1%CuO/MgO/SiO2
catalysts, where they saw an enhanced improvement, with Cu0 as the active sites suggested
by operando X-ray absorption near edge structure (XANES) spectroscopy.106 Deposition-
precipitation (DP) of Au onto MgO/SiO2 catalyst also saw significant increase in the
MgO/SiO2 catalyst.60 DP of AuCl-based precursor surprisingly completely transformed the
bulk MgO into MSH phase. This phenomenon was attributed to the Cl- effect from the Cl-
ion produced from the precursor’s hydrolysis, since acetate precursor did not transform the
MgO into MSH. In-situ titration experiment with propionic acid, a very weak acid, showed
that 1,3-BD production considerably decreased, and did not recover when propionic acid
was not fed anymore. The dehydrogenation reaction to acetaldehyde recovered, however,
indicating the presence of weak basic sites to catalyze dehydrogenation and strong basic
sites to catalyze the subsequent reactions.60 Another transition metals that were used to
23
improve 1,3-BD yield is Ag.59 Silver was dispersed on silica phase when impregnated over
calcined MgO/SiO2 catalyst, as shown by ambient SEM and EDS analysis. Synergistic
effect was observed, where Ag catalyzed the dehydrogenation of ethanol, while MgO-SiO2
phase catalyzed the subsequent reactions. However, increase in Ag loadings seemed to
have created aggregated, larger Ag particles, which in turn deactivated the catalyst.59
Other types of metals had been used as well as promoters. Ohnishi, et al., for
instance, prepared three MgO-SiO2-Na2O and MgO-SiO2-K2O by impregnating aqueous
NaOH and KOH with MgO-SiO2 catalysts prepared from different raw materials using
different methods.90 They found out that the (0.01%) Na2O promoted magnesia-silica
catalyst prepared from ethyl orthosilicate and magnesium nitrate (1:1) by wet-kneading
produced the highest catalytic activity (100%) and selectivity (87%) for formation of 1,3-
BD at 350 ºC for the one-step process.90 The origin of this alkali metal promotion, however,
is not known since there was no characterization reported on how Na and K is coordinated
to the surface. Zirconia, another class of catalyst that will be discussed below, is another
non-redox promoting metal that was used to improve 1,3-BD formation.91 Jones, et al.
found that IWI over an uncalcined MgO/SiO2 gives a higher surface area, since there were
more OH groups to interact with the promoter. Zn itself improved the dehydrogenation
reaction but lagged the aldol condensation, demonstrated by accumulation of acetaldehyde
in the product stream. Co-promotion with zirconia, another solid Lewis acid, was shown
to significantly improve this, since it further provided additional aldol sites on top of the
support’s native sites. The authors suggested that both ZnO and ZrOx were more readily
dispersed over Mg-O-Si linkages, and coprecipitation method was shown to generate more
of this linkage, as shown by 29Si MAS NMR. Furthermore, CHCl3 and NH3 DRIFTS
24
experiments suggested that post-treatment of the catalyst with alkali metals, such as Na,
Li, and K, poisoned the stronger acid sites, without introducing strong basic sites. This
poisoning effect significantly removed ethylene from byproduct, and therefore opening
possibilities of recycling the product back to the system.91
1.2.3. ZrOx –based Catalysts
Study on ethanol to 1,3-BD over ZrOx-based catalysts had been extensively studied
recently. Toussaint, et al.,55 Jones, et al.,78 and Ivanova, et al.,62 screened a number of
different Lewis acid catalysts that would work on aldol condensation and MPV reduction.
The Lewis acid material can then be combined with redox metal catalyst that would
improve the conversion of the dehydrogenation of ethanol. Jones, et al. suggested the use
of ZnO and ZrOx,78 while Ivanova, et al. identified AgO/ZrOx as the most active catalyst.62
The complete reaction network for the catalytic conversion of ethanol into 1,3-BD
over metal-containing (M=Ag, Cu, Ni) oxide catalysts (MOx=MgO, ZrO2, Nb2O5, TiO2,
Al2O3) supported on silica was investigated by Sushkevich, et al.62 From the reaction
network above, these authors narrowed down their catalyst selection to target the
dehydrogenation of ethanol, aldol condensation of acetaldehyde, and MPV-reduction of
the crotonaldehyde. Their experiments demonstrated the superior Ag and Cu-promoted
catalysts over Ni-promoted. Nb and Al-based oxide catalysts developed higher selectivity
toward ethylene due to the BAS, while magnesia and titania oxide catalysts were prominent
in catalyzing aldol condensation. The best performing catalyst was found out to be 1%
Ag/10%ZrO2 on silica reaching 88% conversion and 74% yield of 1,3-BD at 593 K, with
WHSV of 0.04 g g-1 h-1 and time on stream (TOS) of 5 hours.62
25
Optimization of the Ag/ZrOx catalysts was subsequently done by fine-tuning the
composition of both Zn and Zr.61 Catalysts employed were ZrBEA zeolites with various
Zr/Si ratios and zirconia supported on silica, all promoted with silver. Characterization of
the acid sites of the catalysts was conducted by FTIR spectra of deuterated acetonitrile. The
silanol group of the catalysts contributed to the high number or the BAS, giving peaks at
2275 cm-1 that readily disappeared upon evacuation, pointing to the weak interaction of the
OH group. The LAS were also determined using the same technique, particularly by
looking at the band centered at 2303 cm-1. The number of LAS apparently played
significant role in determining which reaction pathways to take place; catalyst with the
highest yield turned out to be the one with the highest LAS. The best catalyst performance
was observed for Ag-promoted ZrBEA (Si/Zr = 100), with 1,3-BD selectivity of 56% and
conversion of 48% under the following conditions: T = 593 K, WHSV = 0.32 g g−1 h−1,
observed after TOS = 3 h.61
DFT calculation, in combination with CO-DRIFTS study, elucidated the nature of
the LAS on the catalyst.107 Two active sites, open and closed isolated LAS, were shown to
be available on the surface. The open site, where there is one terminal hydroxyl group
coordinated to Zr (HO-Zr-(OSi)3), was found to be the main catalytic active sites.107 Over
series of synthesized catalyst, direct correlation was made between the 1,3-BD productivity
and the relative amount of open LAS. The reaction mechanism over this catalyst was
further probed using combination of SSITKA and isotope labeling.108 As expected, two
sites were available for the reaction, the silver sites that were responsible for ethanol
dehydrogenation, and Zr LAS that catalyzed the subsequent aldol and MPV reactions.
Deuterated ethanol CH3CD2OH and CH3CH2OD showed that ethanol dissociatively
26
adsorbed on silanol site, followed by C-H bond breaking over Ag/Si-OH site. The resulting
acetaldehyde was then desorbed from the surface.109 Enolization of acetaldehyde was made
possible by Zr LAS and Zr-O pair sites, and reaction with vapor-phase acetaldehyde
molecule was carried out according to Eley-Rideal mechanism. Facile dehydration of
acetaldol then followed, where crotonaldehyde was produced. MPV reduction by ethanol
was then determined to follow the Langmuir-Hinshelwood mechanism by fitting of kinetic
data.108
Combination of Zn and Zr for ETB reaction had also been explored.93,110 Similar to
previously published endeavors, Zn/Zr mixed oxides were used. From IR-pyridine probing
experiment, it was found that the acidity of the catalyst as a function of Zn/Zr ratio had a
maximum at Zn/Zr = 1:10. The use of Na as a promoter to control the surface acid-base
properties was confirmed by NH3-TPD and IR-Py as well. When 2000 ppm if Na was used
as a promoter, the balanced basic sites and weak BAS exhibited selectivity of 47% at 97%
conversion, with higher productivity per grams of catalyst. Another alkali metal oxide,
cesium oxide, was studied as well as a promoter that eliminates acid sites and reduces
ethylene formation.110 The similarity between Zn/Zr and Ag/Zr catalysts suggested similar
reaction mechanism and active sites. Zn/Zr catalysts, however, required a third promoter
in the form of very small amount of alkali metal to eliminate the acidity it provided, unlike
Ag/Zr catalyst, which did not provide improved Lewis acidity.107
1.2.4. Other Catalysts
This subsection will discuss other metal oxide catalysts that had been
used/investigated in the past. Discussion will cover some of the catalysts that have similar
27
characteristics with Zn(Ag)/Zr catalyst, i.e. Lewis acid combined with dehydrogenation
catalysts, such Hf-based catalyst79 and Ta-based catalyst.82,95 Due to the extensive work
carried out by Sushkevich and Ivanova, Zr-based catalysts were classified as a different
class of catalyst.43,61,62,107–109
Bhattacharyya and Ganguly performed one-step catalytic conversion of ethanol to
1,3-BD in a fixed bed reactor on various single oxide catalysts of aluminum, thorium,
magnesium, iron (III), and zirconium.111 The maximum process yield of 1,3-BD was
achieved at 36.1% using thorium oxide prepared from thorium nitrate and ammonia
carbonate with flow rate at 1.256 mL/g h-1 at 450 ºC. All metal oxides were selected based
on their capabilities of catalyzing both dehydration and dehydrogenation reactions.111 Few
years later, Bhattacharyya and Avasthi conducted exhaustive experiments on the
conversion of ethanol to 1,3-BD via the one-step process on single alumina oxide and
binary alumina-zinc/calcium/chromium/magnesium oxide catalysts. They reported
maximum yield of 72.8% on a fluidized bed reactor using the binary alumina-zinc oxide
catalyst (60:40) as opposed to 55.8% yield on a fixed bed reactor at 425 ºC.85
Work on hafnium oxide catalyst, promoted with Cu and Zn as dehydrogenation
promoter, was done by Baerdemaeker, et al.79 Building on Jones’ catalyst screening,78
slight modification was performed to understand which element contribute to which step
of the mechanisms. Switching the precursors from nitrate to chloride already contributed
in higher 1,3-BD selectivity, stability, and lower ethylene selectivity. Replacing zirconium
with hafnium, a softer material but still in the same group as Zr, suppressed the formation
of ethylene more significantly. Their study also revealed Zn’s superiority in catalyzing
dehydrogenation of ethanol, since when Zn single oxide was used, acetaldehyde
28
accumulated without significant formation of 1,3-BD. It was inferred that, the addition of
Hf suppressed ethanol dehydration, while also promoted the C-C coupling step, i.e. aldol
condensation.
Tantalum oxide, a rare transition metal material, was among the first catalysts used
to catalyze the reaction in the two-step process.68,81 Its use had been revisited following the
renewed interest in the Lebedev process.82,95,112–114 Kyriienko, et al., synthesized a
TaSiBEA zeolite material by dealuminating the starting AlSiBEA with HNO3, followed by
the introduction of tantalum into the framework. 95 The tantalum itself was present as
isolated mononuclear in the framework, as confirmed by UV-Vis and XRD. The catalyst
was further promoted with Ag, Cu, or Zn to introduce dehydrogenation sites. UV-Vis
characterization suggested that Ag was present as Ag(I) and oxidized silver cluster, Cu was
available as isolated mononuclear and oxidized cluster, while Zn was suggested to be in
the framework and as a polynuclear zinc oxide in the extra-framework position.95
Incorporation of Cu and Zn is suggested by pyridine DRIFTS to give higher amount of
LAS, as compared to Ag promotion, while ZnO incorporation exhibited higher BAS
content in the catalyst than the rest. Further deuterated chloroform DRIFTS also suggested
that incorporation of the transition metals into the TaSiBEA catalysts had eliminated
medium strength basic sites, replacing them with weaker ones.95 Mechanistic study for Ta-
SiBEA catalyst,82 and subsequently on Ag-Zr-BEA,114 was done by Müller, et al.82,114
Operando modulated DRIFTS-MS was used to control the ratio of ethanol to acetaldehyde,
maintaining sufficient ethoxy coverage to keep replenishing acetaldehyde formation by
both dehydrogenation and MPV reduction.114 The presence of Ag was reduced
nanoparticles enabled reduction of crotonaldehyde in two possible mechanisms, direct
29
hydrogenation by hydrogen molecule under low ethanol pressure, and MPV reduction
under high ethanol pressure.114
2. Approach
2.1. Approach
The recent emergence of combined dehydrogenating/Lewis acid catalysts has
opened an alternative for the Lebedev’s process. Despite this, the traditional MgO/SiO2
catalysts are still very interesting, due to the lack of suitable characterization methods. Mg
to Si ratio and other synthesis parameters for the available preparation methods are the
most enticing part of the research. Additionally, the catalytic system is still not well-
understood, despite attempts to characterize the system. This is mainly due to the nature of
the catalyst itself that limits the use of several in-situ (operando) spectroscopic methods.
In this study, MgO/SiO2 catalysts were used as the investigated catalysts, where reaction
mechanisms, role of promoters, and catalyst characterization are integral parts of this
dissertation. The approach taken to achieve these objectives consisted of the following
steps:
1. Perform DFT calculation as a preliminary study on proposed reaction mechanisms over
an ideal model MgO surface.
2. Synthesize MgO/SiO2 catalyst with wet-kneading method, fixing Mg to Si ratio to 1
with varying calcination temperature at 500 and 800 °C.
3. Perform reactivity study and reaction mechanism study, both surface and vapor-phase
intermediates.
30
4. Intensively characterize the catalyst using probe molecules and in-situ spectroscopy,
correlate the catalytic active sites with activity and selectivity.
5. Promote the MgO/SiO2 catalyst with transition metal (Cu and Zn), characterize the
system, and investigate the catalytic consequences.
DFT calculation was performed using VASP software package, with MgO (100)
surface termination chosen as the model catalyst. Specifically, a kink site, Mg2+3C O
2-4C pair
were chosen as the active site, since defects had been suggested to have a very high
catalytic activities.115 This first approximation simplified the current state-of-the-art, where
Lewis acidity and basicity were represented by the lower coordinated ionic pairs on the
surface. In reality, the Lewis acidity and basicity were suggested to be provided by the Mg-
O-Si linkages, and promotional effect on the redox site was provided by transition metal
sites on the catalyst.
Experimentally, MgO/SiO2 catalysts were synthesized by wet-kneading in water.
Mg(OH)2 from hydrothermal-NaOH assisted precipitation of magnesium nitrate precursor
and Cab-O-Sil EH5 fumed silica were used as a starting material. Wet-kneading was done
for 4 hours, before separation, drying, and impregnation/calcination. All final catalysts
were calcined at elevated temperature to transform the material into oxides. Experimental
study comprised in-situ and operando spectroscopy, kinetic experiment using fixed bed,
and other bulk and surface ambient characterization such as TEM, SEM, XPS, and XRD.
Materials used were ethanol in inert, such as N2, He, and Ar. Probe molecules were used
to investigate the reaction mechanism, acidity, and basicity of the catalyst.
31
2.2. DFT Calculation
Theoretical (computational) chemistry methods provided an alternative to predict
reaction mechanisms, thermodynamics, kinetics, and even catalyst design by creating
potential energy landscape. The potential energy surface (PES) was created by
optimization of different minima and maxima (TS). This iterative step is essentially solving
the Schrödinger’s equation,
ĤΨ(R,r)=EΨ(R,r) (1.2)
The Hamiltonian in this equation can be separated into two terms, one to account for the
electron and the other for the nucleus. This separation is made possible by using the Born–
Oppenheimer approximation.116 Consequently, the wavefunction is represented as a
product of the electronic and nuclear wavefunctions. Further simplification is also made
by ignoring the nuclear kinetic energy and only taking into account the nuclear’s potential
energy and electron’s kinetic energy. Ab-initio methods and DFT study aim to solve the
Schrödinger’s equation by using electronic wavefunctions in the forms of single electron
molecular orbitals.116
DFT is less computationally expensive than post-HF calculations. It uses electron
(probability) density, ρ(r) to compute the electronic energy. A functional is used to
represent the energy, because energy is a function of electron density, which is a function
of position. This electron density is represented by a sum of one electron orbitals in the
Kohn-Sham equation, 𝜌 = ∑ 𝜑𝑖(𝑟)𝑁𝑖 . The DFT energy is further calculated in the Kohn-
Sham equation,
𝐸[𝜌] = 𝑇𝑠[𝜌] + ∫ 𝑑𝒓 𝑣𝑒𝑥𝑡(𝒓)𝜌(𝒓) + 𝐸𝐻[𝜌] + 𝐸𝑋𝐶[𝜌] (1.3)
32
The terms on the right-hand side represent the (electronic) kinetic energy, external potential
acting on the interacting system, Hartree (or Coulomb) repulsion energy, and the exchange-
correlation energy.116 The exchange-correlation energy is approximated, and several
approximations have been developed, such as the local density approximation (LDA), local
spin density approximation (LSDA), and generalized gradient approximations (GGA). The
generalized gradient approximation (GGA) is an adaptation of the LDA that accounts for
inhomogeneity of the electron density, in which the non-local correction of the gradient of
the electron density (moving away from the coordinate) is added to the exchange-
correlation energy.116 In this work, GGA was used to approximate the exchange-correlation
energy, specifically the simplified Perdew-Burke-Ernzerhof (PBE) GGA. More details of
the computational calculation are provided in Chapter 2
2.3. In-situ and Operando Spectroscopy
In-situ spectroscopy is defined as spectroscopic characterization under relevant,
operating condition.117 The dynamics of the reaction and catalysts at different temperature
can be studied by manipulating the temperature and pressure in a controlled manner.
Operando spectroscopy, on the other hand, is essentially in-situ spectroscopy with
monitoring of the vapor-phase product identification.118,119 Typical setup for operando
spectroscopy comprises a reaction cell with a temperature and/or pressure controller. The
use of this reaction cell enables probing using optical light source, such as Raman and FTIR
spectroscopy, at reaction condition, with flowing gas. The output of this reaction cell is
then connected to product identification system, either GC-MS or MS. Figure 1.3 shows a
typical setup for operando spectroscopy.
33
2.3.1. Infrared Spectroscopy
Infrared (IR) spectroscopy is based on the radiation absorption by the characterized
compounds. When an IR beam is passed through a sample, the compound absorbs radiation
at a specific wavelength, resulting in a decrease of the transmitted radiation. In the spectra,
the absorption is reflected as a dip in transmission, or a peak in absorbance. Upon absorbing
the IR radiation, the molecules’ oscillation amplitude will increase, hence the molecule is
excited to a higher vibrational state. Vibrational states are quantized energy levels, and the
particular wavelength of absorption by a specific bond depends on the energy difference
between the ground level and the excited state. Hence, different bonds produce different
peaks/dips, which also vary depending on the different oscillation modes, in the IR spectra
at different wavenumbers.
Monitoring the molecular events taking place at a certain reaction temperature or
pressure, i.e. in situ IR spectroscopy, is now possible and widely used. A substrate, or
Figure 1.3. Operando spectroscopy setup, flow reaction cell temperature/pressure
controller equipped with FTIR, UV-Vis and Raman spectroscopy that enables real-time
online measurement. Output is connected to real-time GC/MS system. Adapted from: http://www.lehigh.edu/operando
34
catalyst, can be put into a reaction chamber, where a reactant is flown over it. The
spectroscopy is then used to continuously monitor the surface species present on the
substrate while the surface is heated up to a certain desired temperature. This
characterization method is very powerful because one can get information on the bonds
created or destroyed during the temperature ramping, which, when properly analyzed,
could give information on the possible transition states during the reaction.
Vibration spectroscopic measurements of the solid/gas interface can be effectively
carried out using reaction chamber equipped with optical windows enabling measurements
in transmission mode, attenuated total reflectance (ATR) and total or diffuse reflectance
modes. Fourier transform infrared spectroscopy (FTIR) is generally limited to powdered
(submicron-sized) materials, given limitations in throughput.
2.3.2. UV-Vis-NIR Spectroscopy
UV-Vis-NIR spectroscopy is another powerful tool to study catalysts. UV, Vis
(visible), NIR (near infrared) regions cover 200-400 nm, 400-800 nm, and 800-2500 nm
wavelength, respectively.120,121 The UV-Vis region is very useful since it probes the
electronic transitions, while NIR can discover overtones and combination bands of
fundamental vibrational vibrations. This spectroscopy technique is especially useful when
studying the transition metal ions, such as Cu and Zn in this dissertation study. In particular,
two types of bands can be identified in the spectra, charge transfer (CT) bands and d-d
transition bands. CT bands are usually associated with the highest oxidation state, while d-
d transition bands represent reduced states. Bands associated with plasmon resonance are
also available at higher wavelength region, which signifies the presence of a metallic state.
35
Coordination number of the transition metal species can be determined as well
using UV-Vis spectroscopy.122 This is done by comparing the optical absorption edge
energies or band gaps in metal compounds with properties resembling crystalline or
amorphous semiconductors. Several references with known coordination numbers, i.e.
isolated, dimer, trimer, are tested and plotted against the band gap. The band gaps
themselves are obtained from the hν-intercept value [(αhν)1/η = 0] by extrapolating a
straight line from the linear region near the edge on the Tauc plot, where α is absorption
coefficient, hν is the energy of the incident photon, E0 is the optical absorption edge energy,
and for crystalline semiconductors, η is 0.5, 1, 2, and 3, when the optical transitions caused
by photon absorption are direct-allowed, direct-forbidden, indirect-allowed, and indirect-
forbidden, respectively, whereas for amorphous, homogeneous semiconductors, η is
typically 2.122
2.3.3. Operando XANES and EXAFS
High energy X-ray spectroscopy comprises X-ray absorption near edge
spectroscopy (XANES) and extended X-ray absorption fine structure (EXAFS). Sample is
irradiated with a tunable source of high intensity monochromatic x-rays from a synchrotron
radiation facility. The bulk of the solid is penetrated by the x-ray, and only materials with
nearly all the atoms on the surface give surface information. Oxidation state and
coordination information can be extracted using this characterization method. The
characterization can also be done in operando mode, which allows correlation with
catalytic activity.
Not only does XANES give information for oxidation states, but it also
discriminates tetrahedral and octahedral coordination of metal oxides by identification of
36
the pre-edge features. Octahedral coordination has a center of symmetry (unlike
tetrahedral), and transitions between states of g symmetry are not dipole allowed. Hence,
octahedral coordination will not exhibit a pre-edge feature. EXAFS, on the other hand,
provides information on the bond length and coordination number of different atoms on
the molecules. With the correct reference, interpretation can be made to correlate the
change in the bond length, oxidation states, coordination number, or change of coordination
type during the reaction.
2.3.4. Temperature-Programmed Reaction Spectroscopy
Temperature-Programmed reaction spectroscopy (TPRS), often called TPD
(temperature-programmed desorption), TDS (thermal desorption spectroscopy), and TPSR
(temperature-programmed surface reaction), is a very powerful method to characterize a
catalyst.123 When only desorption is taking place, it is either called TDS or TPD, and when
there is a reaction involved, i.e. decomposition, reduction, oxidation, it is called TPRS or
TPSR. This technique requires a very simple infrastructure; a tube with a furnace
containing the catalyst sample and mass spectrometer (MS) that is equipped with vacuum
pump. Typically, the heating rate applied is constant, and hence, partial pressure of the
vapor-phase can be related to the desorption (reaction) temperature. Redhead had shown
that for fast pump rate, the desorption rate is proportional to the pressure in the chamber.124
The desorption kinetics can also be determined, with the desorption activation energy
estimated from the peak temperature. For the first-order kinetic, the following equation is
applicable:
𝐸𝑎
𝑅𝑇𝑝2 =
𝜐
𝛽𝑒
(−𝐸𝑎
𝑅𝑇𝑝) (1.4)
37
In this study, this method was used to identify the intermediates desorbed from the
surface, further confirming the reaction mechanism and rate-determining steps. Ethanol or
other reactants were adsorbed on the surface, flushed to remove the physisorbed species,
and finally temperature was ramped under reactant flow or under inert flow. Depending on
the reaction mechanism, activation energy can be determined, as discussed by Redhead.
Deconvolution of the curve of a m/z number is particularly important. A m/z number can
represent more than one molecules. For instance, m/z = 31 could originate from 1,3-
butanediol, n-butanol, 2-butanol, crotyl alcohol, ethanol, and diethyl ether. Therefore, it is
very important to deconvolute the curve and to consider the important m/z and discard the
others.
2.4. High Sensitivity Low Energy Ion Scattering Spectroscopy (HS-LEIS)
High Sensitivity-Low Energy Ion Scattering (HS-LEIS), was used to probe the
outermost layer of the catalyst, since reaction takes place on the catalyst surface.
Quantitative, elemental information of the outermost layer is probed by bombarding the
surface perpendicularly with noble gas ions. The noble gas ions, which are lighter than the
elements on the surface, will be backscattered, recorded, and analyzed based on the classic
laws of mechanics (conservation of momentum and conservation of energy).125 The
equation can be seen below:
E𝑓 = k2 (𝑚2
𝑚1θ) 𝐸𝑖 (1.5)
Ei is the initial energy carried by the noble gas ion, m1, θ is the backscatter angle, m2 is
the scattering surface atom, k is a known function of m2/m1 and θ, and Ef is the final energy
of the scattered atom that is analyzed by the analyzer. The m2 can hence be determined
from the equation and the element can be determined.125 Elemental composition calculation
38
can further be estimated using quantum chemical method, by first deconvoluting the curve
using Gaussian fitting and integrating the area under the curve to get Ii and use the
following formula:
C𝑖 =
𝐼𝑖
√𝑚𝑖⁄
∑ 𝐼𝑖
√𝑚𝑖⁄
(1.6)
2.5. Probe Molecules
Characterization of (supported) metal oxide catalysts often involves the use of
probe molecules. Probe molecules, combined with spectroscopic techniques, such as
DRIFTS and TPD, can provide a lot of insights on the catalysts surface. This
characterization technique can be conveniently done at relevant reaction temperatures,
which further provides the most relevant information regarding the chemical nature of the
catalyst. In the case of bifunctional catalysts such as MgO/SiO2 catalysts, acidic and basic
probe molecules are indispensable for the acid-base characterization. Two probe molecules
are typically used to characterize the acid sites of a catalysts: ammonia and pyridine.
Ammonia is the most common probe molecule, since it is a very small molecule with
relatively strong basic characteristic, allowing it to penetrate the small pore of the catalysts.
Furthermore, ammonia adsorbs on both BAS as ammonium ion NH4+
and as NH3 by
donating its excess electron pairs to LAS. DRIFTS spectroscopy can be used to
qualitatively and semi-quantitatively determine the acidity of the catalyst. Pyridine, on the
other hand, is a weak but bigger basic molecule. This weak base can differentiate between
LAS and BAS, and is more sensitive to the strong acid sites.
The basicity of the catalysts were probed with various acid probe molecules. CO2
remains the most popular probe molecules, due to its slightly acidic nature and its
39
versatility to bond with basic sites with different strengths. Monodentate, bidentate,
symmetrical, and polydentate (bridged) carbonates are readily formed on the surface with
different splitting values, while bicarbonate can also help identify the presence of weaker
basic sites.126 Methanol can also be used as a reactive probe molecule. On metal oxides,
methanol adsorbs both as a Lewis-bound molecule and as a methoxy species. Upon thermal
treatment, these two species desorb as both molecular methanol and as its reaction product,
depending on which sites it is adsorbed to. On acid sites, methanol undergoes dehydration
to yield dimethyl ether, while on basic sites, CO2 will be produced by consecutive C-H
bonds breaking. Redox site, when available, will transform methanol into formaldehyde.
This versatility is very important when considering a catalyst that possesses different
functional active sites.
2.6. Product Determination with GC-MS/FID
Gas chromatography (GC)-mass spectrometry (MS)/flame ionization detector
(FID) is a very important tool to identify volatile chemical products. The setup essentially
consists as a gas chromatography, coupled with MS or FID as the detector for product
analysis. A GC is an oven box containing capillary or packed column that is internally
coated with a polar solute, which is defines as stationary phase. The oven box will
controllably heat up the column for separation purpose. On top of the oven, there is a
sample loop that functions as a sample storage. At one state, the GC valve will allow sample
to be continuously flown to the sample loop for a storing purpose. At another state, the GC
valve will allow carrier gas, or mobile phase, to flush the sample loop into the GC column
for product identification purpose.
40
When the carrier gas introduces sample to the GC, molecules will be adsorbed on
the stationary phase, while the mobile phase keep flowing through the column. Separation
is based on two variables, the affinity of the molecules (based on polarity) to the column
and boiling point of each molecules. Separation is aided by the heating oven which will
desorb each molecule based on their boiling point. Separation of molecules with similar
boiling points will be based on the affinity to the column. The separated, desorbed
molecules will be carried to the detector, MS and/or FID. Mass spectrometry detects
volatile products by ionizing them and analyzing their mass-to-charge ratio, which is very
specific to different molecules. The ionized molecules are then accelerated to the same
kinetic energy by charged plates down the MS. The ions are deflected using a magnetic
field selectively allowing one m/z to hit the detector. Simultaneous product identification
is made possible by rapidly changing the magnetic field. However, MS is not inherently
quantitative.127
For quantification purpose, FID detector was used in this work. FID detects the ions
formed during combustion of organic compounds in a hydrogen flame.126 These ions are
proportional to the concentration of organic species in the sample. Since this detection
method is based on combustion, the detectable molecules are limited to organic molecules
with C-H bonds. Unlike MS, where the peaks can be identified as a specific molecule, there
is no qualitative way of identifying a single molecule without the use of a standard
molecule. Standard molecules can be used to determine the order of the GC retention time,
and then calibrated to the area recorded by the GC-FID, and response factor for each
molecule can be determined as the area/concentration.
41
3. Thesis Outline
Chapter 2
The complete experimental methods were presented in this chapter, which covers
all experiments/computation done in this work. Results and discussions were presented in
the following chapter accordingly.
Chapter 3
This chapter discusses the study of relevant reaction mechanisms by DFT
simulation. Three previously proposed reaction mechanisms were investigated over a kink
site of MgO (100) crystal model catalyst. This study was represented by comparison of
kinetics and thermodynamics of the three reaction mechanisms, demonstrated by plotted
potential energy surfaces. This work was published in J. Catal. 2017, 346, 78–91.
Chapter 4
This chapter follows up the mechanistic study using in-situ DRIFTS to investigate
the surface reaction mechanism on the surface of MgO/SiO2 catalyst. Detailed study using
different feed, such as crotonaldehyde/ethanol, acetaldehyde, ethanol/inert, and
ethanol/ethanol, were used in combination with GC-MS to identify the product stream.
This work was published in Catal. Sci. Technol. 2017, 7 (20), 4648–4668.
Chapter 5
This chapter discusses the catalyst characterization of MgO/SiO2, as well as the
kinetics of the reaction. Identification of the active sites were done by titration of the
catalyst, as studied using DRIFTS and fixed-bed reactor with GC-MS. Characterization
was done using XRD, HS-LEIS, and probe molecules. This work was submitted to Journal
of Physical Chemistry C journal.
42
Chapter 6
The catalyst was promoted with transition metal, i.e. Zn and Cu. Initial
computational study using a first approximation for the catalyst model was done to
compare the reaction mechanisms on promoted catalysts. Promotional effect was briefly
discussed by comparison of reactivity data, which is followed by characterization using
UV-Vis spectroscopy, TEM and DRIFTS of CO2 and NH3. Subsequently, the chemical
nature of the catalyst was studied by operando methanol DRIFTS to probe the chemical
change of the catalyst brought about by promoting the catalyst with transition metals. The
catalysts were further studied for its reaction. In-situ DRIFTS and cofeeding with different
poison probe molecules were carried out to study the active sites of the catalyst. This study
was completed by operando XANES/EXAFS to investigate the chemical change of the
promoter materials during reaction. Part of this work was submitted to ACS Catalysis
journal.
Chapter 7
A summary of the whole work for this dissertation is presented. Future outlook was
discussed as well, presenting works that need to be further done.
43
References
(1) McCoy, M.; Reisch, M. S.; Tullo, A. H.; Short, P. L.; Tremblay, J. F.; Storck, W.
J. Chem. Eng. News 2007, 85, 29.
(2) Werpy, T.; Petersen, G. Top Value Added Chemicals from Biomass. Volume 1.
Results of Screening for Potential Candidates from Sugars and Synthesis Gas; US
Department of Energy: Washington DC, 2004.
(3) Keturakis, C. Operando Molecular Spectroscopy during Catalytic Biomass
Pyrolysis, Lehigh University, 2015.
(4) Angelici, C.; Weckhuysen, B. M.; Bruijnincx, P. C. a. ChemSusChem 2013, 6 (9),
1595–1614.
(5) Sun, J.; Wang, Y. ACS Catal. 2014, 4 (4), 1078–1090.
(6) Ni, M.; Leung, D. Y. C.; Leung, M. K. H. Int. J. Hydrogen Energy 2007, 32 (15),
3238–3247.
(7) Haryanto, A.; Fernando, S.; Murali, N.; Adhikari, S. Energy & Fuels 2005, 19 (5),
2098–2106.
(8) Hanspal, S.; Young, Z. D.; Shou, H.; Davis, R. J. ACS Catal. 2015, 1737–1746.
(9) Morrill, M. R.; Thao, N. T.; Shou, H.; Davis, R. J.; Barton, D. G.; Ferrari, D.;
Agrawal, P. K.; Jones, C. W. ACS Catal. 2013, 3 (7), 1665–1675.
(10) Birky, T. W.; Kozlowski, J. T.; Davis, R. J. J. Catal. 2013, 298 (0), 130–137.
(11) Kulkarni, N. V; Brennessel, W. W.; Jones, W. D. ACS Catal. 2017, 997–1002.
(12) Young, Z. D.; Hanspal, S.; Davis, R. J. ACS Catal. 2016, 6 (5), 3193–3202.
(13) Ho, C. R.; Shylesh, S.; Bell, A. T. ACS Catal. 2016, 6 (2), 939–948.
(14) Takahara, I.; Saito, M.; Inaba, M.; Murata, K. Catal. Letters 2005, 105 (3), 249–
252.
(15) Bi, J.; Guo, X.; Liu, M.; Wang, X. Catal. Today 2010, 149 (1), 143–147.
(16) Phung, T. K.; Busca, G. Catal. Commun. 2015, 68, 110–115.
(17) Zhang, X.; Wang, R.; Yang, X.; Zhang, F. Microporous Mesoporous Mater. 2008,
116 (1–3), 210–215.
(18) Han, Y.; Lu, C.; Xu, D.; Zhang, Y.; Hu, Y.; Huang, H. Appl. Catal. A Gen. 2011,
396 (1–2), 8–13.
(19) DeWilde, J. F.; Chiang, H.; Hickman, D. A.; Ho, C. R.; Bhan, A. ACS Catal. 2013,
3 (4), 798–807.
(20) Phung, T. K.; Proietti Hernández, L.; Lagazzo, A.; Busca, G. Appl. Catal. A Gen.
2015, 493, 77–89.
(21) Gayubo, A. G.; Tarrío, A. M.; Aguayo, A. T.; Olazar, M.; Bilbao, J. Ind. Eng.
Chem. Res. 2001, 40 (16), 3467–3474.
(22) Gayubo, A. G.; Alonso, A.; Valle, B.; Aguayo, A. T.; Olazar, M.; Bilbao, J. Fuel
2010, 89 (11), 3365–3372.
(23) Iwamoto, M. Molecules 2011, 16 (9), 7844–7863.
(24) Meng, T.; Mao, D.; Guo, Q.; Lu, G. Catal. Commun. 2012, 21 (0), 52–57.
(25) Song, Z.; Takahashi, A.; Nakamura, I.; Fujitani, T. Appl. Catal. A Gen. 2010, 384
(1–2), 201–205.
(26) Liu, C.; Sun, J.; Smith, C.; Wang, Y. Appl. Catal. A Gen. 2013, 467 (0), 91–97.
(27) Sun, J.; Zhu, K.; Gao, F.; Wang, C.; Liu, J.; Peden, C. H. F.; Wang, Y. J. Am.
Chem. Soc. 2011, 133 (29), 11096–11099.
44
(28) Sun, J.; Baylon, R. A. L.; Liu, C.; Mei, D.; Martin, K. J.; Venkitasubramanian, P.;
Wang, Y. J. Am. Chem. Soc. 2016, 138 (2), 507–517.
(29) Lippits, M. J.; Nieuwenhuys, B. E. Catal. Today 2010, 154 (1), 127–132.
(30) Lippits, M. J.; Nieuwenhuys, B. E. J. Catal. 2010, 274 (2), 142–149.
(31) Tu, Y.-J.; Chen, Y.-W. Ind. Eng. Chem. Res. 1998, 37 (7), 2618–2622.
(32) Tu, Y.-J.; Li, C.; Chen, Y.-W. J. Chem. Technol. Biotechnol. 1994, 59 (2), 141–
147.
(33) Tu, Y.-J.; Chen, Y.-W.; Li, C. J. Mol. Catal. 1994, 89 (1), 179–189.
(34) Chang, F.-W.; Yang, H.-C.; Roselin, L. S.; Kuo, W.-Y. Appl. Catal. A Gen. 2006,
304 (0), 30–39.
(35) Chang, F.-W.; Kuo, W.-Y.; Lee, K.-C. Appl. Catal. A Gen. 2003, 246 (2), 253–
264.
(36) Gaspar, A. B.; Barbosa, F. G.; Letichevsky, S.; Appel, L. G. Appl. Catal. A Gen.
2010, 380 (1–2), 113–117.
(37) Inui, K.; Kurabayashi, T.; Sato, S. Appl. Catal. A Gen. 2002, 237 (1–2), 53–61.
(38) Inui, K.; Kurabayashi, T.; Sato, S.; Ichikawa, N. J. Mol. Catal. A Chem. 2004, 216
(1), 147–156.
(39) Sánchez, A. B.; Homs, N.; Fierro, J. L. G.; Piscina, P. R. de la. Catal. Today 2005,
107–108 (0), 431–435.
(40) Jørgensen, B.; Egholm Christiansen, S.; Dahl Thomsen, M. L.; Christensen, C. H.
J. Catal. 2007, 251 (2), 332–337.
(41) Santacesaria, E.; Carotenuto, G.; Tesser, R.; Di Serio, M. Chem. Eng. J. 2012, 179
(0), 209–220.
(42) Inui, T.; Matsuda, H.; Yamase, O.; Nagata, H.; Fukuda, K.; Ukawa, T.; Miyamoto,
A. J. Catal. 1986, 98 (2), 491–501.
(43) Palagin, D.; Sushkevich, V. L.; Ivanova, I. I. J. Phys. Chem. C 2016, 120 (41),
23566–23575.
(44) Chalk, S. G.; Miller, J. F. J. Power Sources 2006, 159 (1), 73–80.
(45) Taifan, W.; Baltrusaitis, J. Appl. Catal. B Environ. 2016, 198, 525–547.
(46) Ni, M.; Leung, M. K. H.; Leung, D. Y. C.; Sumathy, K. Renew. Sustain. Energy
Rev. 2007, 11 (3), 401–425.
(47) Matar, S.; Hatch, L. F. F. Chemistry of Petrochemical Processes; Elsevier, 2001.
(48) Yang, C.; Meng, Z. Y. J. Catal. 1993, 142 (1), 37–44.
(49) Kozlowski, J. T.; Davis, R. J. J. Energy Chem. 2013, 22 (1), 58–64.
(50) León, M.; Díaz, E.; Ordóñez, S. Catal. Today 2011, 164 (1), 436–442.
(51) Carvalho, D. L.; de Avillez, R. R.; Rodrigues, M. T.; Borges, L. E. P.; Appel, L. G.
Appl. Catal. A Gen. 2012, 415–416 (0), 96–100.
(52) Marcu, I.-C.; Tichit, D.; Fajula, F.; Tanchoux, N. Catal. Today 2009, 147 (3–4),
231–238.
(53) Tsuchida, T.; Kubo, J.; Yoshioka, T.; Sakuma, S.; Takeguchi, T.; Ueda, W. J.
Catal. 2008, 259 (2), 183–189.
(54) Quattlebaum, W. M.; Toussaint, W. J.; Dunn, J. T. J. Am. Chem. Soc. 1947, 69,
593–599.
(55) Toussaint, W. J.; Dunn, J. T.; Jachson, D. R. Ind. Eng. Chem. 1947, 39 (2), 120–
125.
(56) Lebedev, S. V. Zh Obs. Khim 1933, No. 3, 698–717.
45
(57) Larina, O.; Kyriienko, P.; Soloviev, S. Catal. Letters 2015, 1–7.
(58) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A.
ChemSusChem 2014, 7 (9), 2505–2515.
(59) Janssens, W.; Makshina, E. V. V.; Vanelderen, P.; De Clippel, F.; Houthoofd, K.;
Kerkhofs, S.; Martens, J. A. A.; Jacobs, P. A. A.; Sels, B. F. B. F. ChemSusChem
2014, 8 (6), 994–1008.
(60) Shylesh, S.; Gokhale, A. A.; Scown, C. D.; Kim, D.; Ho, C. R.; Bell, A. T.
ChemSusChem 2016, 9 (12), 1462–1472.
(61) Sushkevich, V. L.; Ivanova, I. I.; Taarning, E. Green Chem. 2015.
(62) Sushkevich, V. L.; Ivanova, I. I.; Ordomsky, V. V; Taarning, E. ChemSusChem
2014, 7 (9), 2527–2536.
(63) Taifan, W.; Yan, G. X.; Baltrusaitis, J. Catal. Sci. Technol. 2017, 7 (20), 4648–
4668.
(64) Ordomsky, V. V.; Sushkevich, V. L.; Ivanova, I. I. J. Mol. Catal. A Chem. 2010,
333 (1), 85–93.
(65) Phung, T. K.; Lagazzo, A.; Rivero Crespo, M. Á.; Sánchez Escribano, V.; Busca,
G. J. Catal. 2014, 311, 102–113.
(66) Taifan, W. E.; Bučko, T.; Baltrusaitis, J. J. Catal. 2017, 346, 78–91.
(67) Georgieff, K. K. J. Appl. Polym. Sci. 1966, 10 (9), 1305–1313.
(68) Wang, Z. In Comprehensive Organic Name Reactions and Reagents; John Wiley
& Sons, Inc., 2010.
(69) Ostromislenskiy, I. J. Russ. Phys. Chem. Soc. 1915, No. 47, 1472 –1506.
(70) Gruver, V.; Sun, A.; Fripiat, J. J. Catal. Letters 1995, 34 (3–4), 359–364.
(71) Chieregato, A.; Velasquez Ochoa, J.; Bandinelli, C.; Fornasari, G.; Cavani, F.;
Mella, M. ChemSusChem 2014, 8 (2), 377–388.
(72) Makshina, E. V; Dusselier, M.; Janssens, W.; Degreve, J.; Jacobs, P. A.; Sels, B. F.
Chem. Soc. Rev. 2014, 43 (22), 7917–7953.
(73) Pomalaza, G.; Capron, M.; Ordomsky, V.; Dumeignil, F. Catalysts 2016, 6 (12),
203–237.
(74) Makshina, E. V.; Janssens, W.; Sels, B. F.; Jacobs, P. A. Catal. Today 2012, 198
(1), 338–344.
(75) Niiyama, H.; Morii, S.; Echigoya, E. Bull. Chem. Soc. Jpn. 1972, 45 (3), 655–659.
(76) Chung, S.-H.; Angelici, C.; Hinterding, S. O. M.; Weingarth, M.; Baldus, M.;
Houben, K.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS Catal. 2016, 6 (6),
4034–4045.
(77) Huang, X.; Men, Y.; Wang, J.; An, W.; Wang, Y. Catal. Sci. Technol. 2017, 7 (1),
168–180.
(78) Jones, M. D.; Keir, C. G.; Iulio, C. Di; Robertson, R. A. M.; Williams, C. V;
Apperley, D. C. Catal. Sci. Technol. 2011, 1 (2), 267–272.
(79) De Baerdemaeker, T.; Feyen, M.; Müller, U.; Yilmaz, B.; Xiao, F.-S.; Zhang, W.;
Yokoi, T.; Bao, X.; Gies, H.; De Vos, D. E. ACS Catal. 2015, 3393–3397.
(80) Bhattacharyya, S. K.; Sanyal, S. K. J. Catal. 1967, 7 (2), 152–158.
(81) Corson, B. B.; Jones, H. E.; Welling, C. E.; Hinckley, J. A.; Stahly, E. E. Ind. Eng.
Chem. 1950, 42 (2), 359–373.
(82) Muller, P.; Burt, S. P.; Love, A. M.; McDermott, W. P.; Wolf, P.; Hermans, I. ACS
Catal. 2016, 6 (10), 6823–6832.
46
(83) Jones, M. D. Chem. Cent. J. 2014, 8 (1), 53.
(84) Posada, J. A.; Patel, A. D.; Roes, A.; Blok, K.; Faaij, A. P. C.; Patel, M. K.
Bioresour. Technol. 2013, 135, 490–499.
(85) Bhattacharyya, S. K.; Avasthi, B. N. Ind. Eng. Chem. Process Des. Dev. 1963, 2
(1), 45–51.
(86) Hayashi, Y.; Akiyama, S.; Miyaji, A.; Sekiguchi, Y.; Sakamoto, Y.; Shiga, A.;
Koyama, T.; Motokura, K.; Baba, T. Phys. Chem. Chem. Phys. 2016, 18 (36),
25191–25209.
(87) Velasquez Ochoa, J.; Bandinelli, C.; Vozniuk, O.; Chieregato, A.; Malmusi, A.;
Recchi, C.; Cavani, F. Green Chem. 2016, 18, 1653–1663.
(88) Velasquez Ochoa, J.; Malmusi, A.; Recchi, C.; Cavani, F.; Ochoa, J. V.; Malmusi,
A.; Recchi, C.; Cavani, F. ChemCatChem 2017, 9 (12), 2128–2135.
(89) Larina, O. V.; Kyriienko, P. I.; Trachevskii, V. V.; Vlasenko, N. V.; Soloviev, S.
O. Theor. Exp. Chem. 2016, 51 (6), 387–393.
(90) Ohnishi, R.; Akimoto, T.; Tanabe, K. J. Chem. Soc. Chem. Commun. 1985, No.
22, 1613–1614.
(91) Da Ros, S.; Jones, M. D.; Mattia, D.; Pinto, J. C.; Schwaab, M.; Noronha, F. B.;
Kondrat, S. A.; Clarke, T. C.; Taylor, S. H. ChemCatChem 2016, 8, 1–12.
(92) Sushkevich, V. L.; Ivanova, I. I. ChemSusChem 2016, 9 (16), 2216–2225.
(93) Baylon, R. a. L.; Sun, J.; Wang, Y. Catal. Today 2015, 1–7.
(94) Larina, O. V; Kyriienko, P. I.; Soloviev, S. O. Theor. Exp. Chem. 2016, 52 (1), 51–
56.
(95) Kyriienko, P. I.; Larina, O. V; Soloviev, S. O.; Orlyk, S. M.; Calers, C.; Dzwigaj,
S.; Pisarzhevsky, L. V. ACS Sustain. Chem. Eng. 2017, 5 (3), 2075–2083.
(96) Natta, G.; Rigamonti, R. Chim. E Ind. 1947, 29, 195–201.
(97) Ruckenstein, E.; Khan, A. Z. J. Catal. 1993, 141 (2), 628–647.
(98) Cavalleri, M.; Pelmenschikov, A.; Morosi, G.; Gamba, A.; Coluccia, S.; Martra, G.
In Oxide-based Systems at the Crossroads of ChemistrySecond International
Workshop October 8-11, 2000, Como, Italy; A. Gamba, C. C. and S. C. B. T.-S. in
S. S. and C., Ed.; Elsevier, 2001; Vol. Volume 140, pp 131–139.
(99) Kvisle, S.; Aguero, A.; Sneeden, R. P. A. Appl. Catal. 1988, 43 (1), 117–131.
(100) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A. Catal.
Sci. Technol. 2015.
(101) Natta, G.; Rigamonti, R. Chim. Ind. 1947, 29, 239–243.
(102) Lewandowski, M.; Babu, G. S.; Vezzoli, M.; Jones, M. D.; Owen, R. E.; Mattia,
D.; Plucinski, P.; Mikolajska, E.; Ochenduszko, A.; Apperley, D. C. Catal.
Commun. 2014, 49, 25–28.
(103) Kitayama, Y.; Michishita, A. J. Chem. Soc., Chem. Commun. 1981, No. 9, 401–
402.
(104) Kitayama, Y.; Satoh, M.; Kodama, T. Catal. Letters 1996, 36 (1–2), 95–97.
(105) Sekiguchi, Y.; Akiyama, S.; Urakawa, W.; Koyama, T.; Miyaji, A.; Motokura, K.;
Baba, T. Catal. Commun. 2015, 68, 20–24.
(106) Angelici, C.; Meirer, F.; van der Eerden, A. M. J.; Schaink, H. L.; Goryachev, A.;
Hofmann, J. P.; Hensen, E. J. M.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS
Catal. 2015, 5 (10), 6005–6015.
(107) Sushkevich, V. L.; Palagin, D.; Ivanova, I. I. ACS Catal. 2015, 4833–4836.
47
(108) Sushkevich, V. L.; Ivanova, I. I. Appl. Catal. B Environ. 2017, 215, 36–49.
(109) Sushkevich, V. L.; Ivanova, I. I.; Taarning, E. ChemCatChem 2013, 5 (8), 2367–
2373.
(110) Patil, P. T.; Liu, D.; Liu, Y.; Chang, J.; Borgna, A. Appl. Catal. A Gen. 2017, 543,
67–74.
(111) Bhattacharyya, S. K.; Ganguly, N. D. J. Appl. Chem. 1962, 12 (3), 97–104.
(112) Kim, T.-W.; Kim, J.-W.; Kim, S.-Y.; Chae, H.-J.; Kim, J.-R.; Jeong, S.-Y.; Kim,
C.-U. Chem. Eng. J. 2014.
(113) Chae, H.-J.; Kim, T.-W.; Moon, Y.-K.; Kim, H.-K.; Jeong, K.-E.; Kim, C.-U.;
Jeong, S.-Y. Appl. Catal. B Environ. 2014, 150–151, 596–604.
(114) Müller, P.; Wang, S.-C.; Burt, S. P.; Hermans, I. ChemCatChem 2017, 9 (18),
3572–3582.
(115) Branda, M. M.; Rodríguez, A. H.; Belelli, P. G.; Castellani, N. J. Surf. Sci. 2009,
603 (8), 1093–1098.
(116) Jensen, F. Introduction to Computational Chemistry, 2nd ed.; John Wiley & Sons
Ltd., 2006.
(117) Weckhuysen, B. M. Phys. Chem. Chem. Phys. 2003, 5 (20), 4351–4360.
(118) Chakrabarti, A.; Ford, M. E.; Gregory, D.; Hu, R.; Keturakis, C. J.; Lwin, S.;
Tang, Y.; Yang, Z.; Zhu, M.; Bañares, M. A.; Wachs, I. E. Catal. Today 2017,
283, 27–53.
(119) Bañares, M. A. Catal. Today 2005, 100 (1), 71–77.
(120) Jentoft, F. C. B. T.-A. in C. Academic Press, 2009; Vol. 52, pp 129–211.
(121) Schoonheydt, R. A. Chem. Soc. Rev. 2010, 39 (12), 5051–5066.
(122) Bravo-Suarez, J. J.; Subramaniam, B.; Chaudhari, R. V. J. Phys. Chem. C 2012,
116 (34), 18207–18221.
(123) Gorte, R. J. Catal. Today 1996, 28 (4), 405–414.
(124) Redhead, P. a. Vacuum 1962, 12 (4), 203–211.
(125) ter Veen, H. R. J.; Kim, T.; Wachs, I. E.; Brongersma, H. H. Catal. Today 2009,
140 (3), 197–201.
(126) McWilliam, I. G.; Dewar, R. A. Nature 1958, 181, 760.
(127) Doerr, A. Nat. Methods 2008, 6, 34.
48
Chapter 2
Experimental Methods
1. Introduction ........................................................................................................ 48
2. Computational Details ....................................................................................... 49
2.1. Electronic structure calculations .............................................................. 49
2.2. Structural optimization calculations ......................................................... 49
2.3. Structural model ....................................................................................... 50
2.4. Free-energy calculation ............................................................................ 52
2.4.1. Working equations ................................................................................... 52
2.4.2. Partitioning of atomic degrees of freedom in interacting and non-
interacting systems ............................................................................................... 53
2.4.3. Calculation of harmonic vibrational frequencies ..................................... 54
3. Experimental Methods ...................................................................................... 55
3.1. Catalyst synthesis ....................................................................................... 55
3.1.1. Synthesis of magnesium oxide, MgO, catalyst ......................................... 55 3.1.2. Synthesis of MgO/SiO2 catalysts .............................................................. 55
3.1.3. Synthesis of promoted wet-kneaded MgO/SiO2 catalysts ......................... 56
3.2. Catalytic reactivity study ........................................................................... 56
3.3. Catalyst characterization ............................................................................ 58
3.3.1. High-sensitivity low energy ion scattering (HS-LEIS) .............................. 58
3.3.2. XRD and BET surface area ........................................................................59
3.3.3. Transition metal concentration measurements ........................................... 59
3.3.4. Scanning transmission electron microscopy .............................................. 60
3.3.5. In-situ spectroscopy ................................................................................... 60
3.3.6. Acid-base characterization using pyridine, NH3, CO2, and methanol as
probe molecules .................................................................................................... 61
3.4. Reaction mechanism study using in-situ DRIFTS spectroscopy and TPRS
............................................................................................................................... 62
3.5. Operando XANES and EXAFS spectroscopy during ethanol reaction to
1,3-BD over Cu- and Zn-promoted MgO/SiO2 catalysts ...................................... 64
References ....................................................................................................................... 65
1. Introduction
This chapter is dedicated for both computational details and experimental methods
used throughout the research work.
49
2. Computational Details
2.1. Electronic structure calculations.
Periodic DFT calculations have been performed using the VASP code.1–4 The
Kohn–Sham equations have been solved variationally in a plane-wave basis set using the
projector-augmented-wave (PAW) method of Blochl,5 as adapted by Kresse and Joubert.5
The exchange-correlation energy was described by the PBE generalized gradient
approximation.6 Brillouin-zone was sampled using 2x2x1 k-point mesh. The plane-wave
cutoff was set to 400 eV. The convergence criterion for the electronic self-consistency
cycle, measured by the change in the total energy between successive iterations, was set to
10-6 eV/cell.
2.2. Structural optimization calculations.
Transition states have been identified using the DIMER method,7 as improved by
Heyden et al.8 Atomic positions were considered to be relaxed if all forces acting on the
atoms were smaller than 0.005 eV/Å. Transition states were proven to be first-order saddle
points of the potential energy surface using vibrational analysis. The intrinsic reaction
coordinates9,10 (IRCs) for the forward and backward reaction steps were identified using
the damped velocity Verlet algorithm.11 The structures corresponding to potential energy
minima along the IRC were further relaxed using a conjugate-gradient algorithm such as
to satisfy the same optimization criterion as for transition states. Vibrational analysis was
performed to ensure that the relaxed structures correspond to true potential energy minima.
This procedure guarantees that reactant and product states are linked by a path with a single
transition state.
50
The Gibbs free-energy calculations have been performed using the harmonic/rigid
rotor approximation to the transition state theory12 for the temperature of 723 K. The
detailed calculations for the Gibbs energy estimations are provided in subchapter 2.4. All
relative energies were referenced to the sum of the relaxed MgO slab, and three molecules
of gas phase ethanol. The first-order reaction constants were computed using:
𝑘(𝑇) =𝑘𝐵𝑇
ℎ𝑒−
∆𝐺𝐴(𝑇)
𝑅𝑇 (2.1)
where kB – 1.380662x10-23 m2 kg/s2 K1, T = 723 K, h - 6.626176x10-34 m2 kg/s, R - 1.987
cal/K mol,and GA(T) is molar Gibbs free-energy of activation defined for the forward
and reverse reactions as follows:
𝛥𝐺𝐴,𝑓𝑜𝑟𝑤𝑎𝑟𝑑 = 𝐺𝑇𝑆 − 𝐺𝐼𝑆 (2.2)
𝛥𝐺𝐴,𝑟𝑒𝑣𝑒𝑟𝑠𝑒 = 𝐺𝑇𝑆 − 𝐺𝐹𝑆 (2.3)
with subscripts TS, FS and IS representing transition state, final state, and initial state,
respectively. Similarly, the reaction free energy (𝛥𝐺𝑅𝑥) is defined as follows:
𝛥𝐺𝑅𝑥 = 𝐺𝐹𝑆 − 𝐺𝐼𝑆 (2.4)
The rate constants for the forward and reverse reaction steps are labelled as kf and kr,
respectively.
2.3. Structural model.
Under normal conditions, MgO crystallizes in the rocksalt crystal structure, with
(001) being the most prominent surface. For the specific reaction of ethanol to 1,3-
butadiene, the original working catalyst is MgO/SiO2 13,14. The addition of MgO to SiO2,
regardless of the methods used to synthesize the catalysts, does not result in the formation
of MgSiO4 solid solution.13,15–17 It was found that MgO crystalline phase was the only
phase found on the amorphous surface. In contrast, a study by Angelici et al.18 claims that
51
MgSiO4 phase was indeed formed and it was suggested to be responsible for the aldol
condensation of the acetaldehyde intermediates. The authors argued that FTIR shows the
presence of a hydroxyl group which coincides with that found in talc. This statement
contradicts their previous study, in which the presence of the silicate was not confirmed by
XRD.19 The addition of MgO to amorphous silica combined with increased surface area of
the catalyst is also known to create more defects on the MgO itself, which clearly
contributes to the higher activity and selectivity of the catalyst. The unsaturated surface
oxygen atoms, e.g. corner and edge O, are unfavorable energetically leading to the
tendency to release an electron and to turn into O- species. Hence these unsaturated oxygen
atoms act as donors of an electron pair, i.e. as Lewis bases.20 Cube corners,21 terraces, steps,
corners, and reverse corners were studied extensively.22
Periodic three layer slab of MgO consisting of 8 x 6 x 3 primitive cells was used
throughout all the calculations. Previous investigations have identified defect sites of MgO
catalysts, most notably the three- and four-fold coordinated surface atoms (Mg2+3C O2-
4C),
Figure 2.1 Periodic MgO slab used throughout the calculations. The whited out bottom
layer indicates the atoms whose positions were kept frozen during calculations.
52
as the active sites22–27 whereas atoms at surface terraces were found to be relatively
unreactive.28 The positions of atoms in the bottommost layer were fixed while remaining
atoms were relaxed. PBE lattice parameter of 4.255 Å was used to construct the slab. This
value is similar to values of 4.261 (PBE) and 4.186 Å (experimental).29 In this work we
considered the Mg3C coordinated to O4C in the form of stepped kink, depicted in Figure 2.1,
as the active sites for the reaction. The large slab is used to accommodate several
intermolecular reactions that might include interactions of C2 and C4 intermediates and
also to avoid significant interaction between molecules in the neighboring unit cells.
2.4. Free-energy calculation
The free-energy calculations have been performed using a static approach based on
harmonic and rigid rotor approximations to vibrational and rotational degrees of freedom.
Although this methodology is described in detail in many textbooks,12 we find it useful to
summarize the working equations in this text.
2.4.1. Working equations
Within the static approximation used in this study, Gibbs free energy for each state
is expressed as a sum of contribution of electronic (el), vibrational (vib), rotational (rot),
and translational (tr) degrees of freedom (DOF):
𝐺 = 𝐺𝑒𝑙 + 𝐺𝑣𝑖𝑏 + 𝐺𝑟𝑜𝑡 + 𝐺𝑡𝑟 (2.5)
Discussion of partitioning of atomic degrees of freedom in the case of interacting and non-
interacting systems is provided in section 1.2.
The electronic contribution for a system in a singlet electronic state (all systems discussed
in this work) takes a form:
𝐺𝑒𝑙 = 𝐸𝐷𝐹𝑇 (2.6)
53
where 𝐸𝐷𝐹𝑇 is a Kohn-Sham energy computed by solving Schrödinger equation at the DFT
level. The vibrational contribution is expressed as
𝐺𝑣𝑖𝑏 =3
2𝑘𝐵𝑇 − 𝑇𝑘𝐵 ∑ (
ℎ𝜈𝑖
𝑘𝐵𝑇
1
𝑒
ℎ𝜈𝑖𝑘𝐵𝑇−1
− ln (1 − 𝑒−
ℎ𝜈𝑖𝑘𝐵𝑇))
𝑁𝑣𝑖𝑏𝑖=1 (2.7)
where 𝑘𝐵 and ℎ are fundamental constants (Boltzmann and Planck, respectively), 𝑇 is
thermodynamic temperature (723 K), 𝑁𝑣𝑖𝑏 is the number of vibrational degrees of freedom,
and 𝜈 is a harmonic vibrational frequency. The rotational contribution is expressed as
𝐺𝑟𝑜𝑡 = −𝑘𝐵𝑇 ln (√𝜋
𝜎(
8𝜋2𝑘𝐵𝑇
ℎ2 )3/2
√𝐼1𝐼2𝐼3) (2.8)
where 𝜎 is a symmetry index, and 𝐼1, 𝐼2, and 𝐼3 are moments of inertia of a molecule. The
values of 𝜎 for molecules considered in this study are compiled in Tab.S1. Finally, the term
𝐺𝑡𝑟 is expressed as follows
𝐺𝑡𝑟 = −𝑘𝐵𝑇 ln (𝑉𝑚 (2𝜋𝑀𝑘𝐵𝑇
ℎ2 )3/2
) (2.9)
where 𝑉𝑚 = 𝑘𝐵𝑇/𝑝 is a volume occupied by one particle of ideal gas at given external
pressure 𝑝 (101325 Pa) and temperature (723 K), and 𝑀 is the total molecular mass.
Table 2.1. Symmetry indices 𝜎 for gas-phase molecules considered in this study.
Molecule 𝝈
acetaldehyde 1
crotonaldehyde 1
ethanol 1
butadiene 2
dihydrogen 2
water 2
2.4.2. Partitioning of atomic degrees of freedom in interacting and non-interacting
systems
For a system consisting of 𝑁𝑠,𝑓𝑟𝑒𝑒 free substrate atoms (in our model 𝑁𝑠,𝑓𝑟𝑒𝑒
corresponds to all substrate atoms except of the bottommost layer of the slab, see
54
subchapter 2.3 and Figure 2.1) and 𝑁𝑀 atoms forming molecules adsorbed on the substrate,
the number of vibrational degrees of freedom is 𝑁𝑣𝑖𝑏 = 3(𝑁𝑠,𝑓𝑟𝑒𝑒 + 𝑁𝑀). There are no
rotational and translational degrees of freedom in such a case and hence also the
contribution of 𝐺𝑟𝑜𝑡 and 𝐺𝑡𝑟 to Gibbs free energy is zero. In the case of systems consisting
of a substrate and 𝑃 molecules that neither interact with each other nor they interact with
the substrate, the total number of vibrational degrees of freedom is 3𝑁𝑠,𝑓𝑟𝑒𝑒 +
∑ (3𝑁𝑀𝑖− 3 − 𝑁𝑟𝑜𝑡,𝑖)
𝑃𝑖=1 , where 𝑁𝑀𝑖
and 𝑁𝑟𝑜𝑡,𝑖 are the number of atoms and the number
of rotational degrees of freedom in the molecule i. Each molecule has 3 translational and 2
(linear molecules) or 3 (nonlinear molecules) rotational degrees of freedom.
2.4.3. Calculation of harmonic vibrational frequencies
The harmonic vibrational frequencies have been computed using the finite
differences method implemented in VASP. The numerical differentiation has been done
using the a differences formula with displacement of size 0.02 Å. Even the use of quite
stringent relaxation criterion (maximal force smaller than 0.005 eV/ Å) does not ensure the
correct eigenvalue spectrum of dynamical matrix (i.e. zero imaginary vibrational
frequencies in the case of minima and one imaginary vibrational frequency in the case of
first-order saddle) in all cases. In order to obtain correct vibrational spectrum in such a
problematic case, several iterations consisting of line-minimization of energy along the
incorrect unstable directions, followed by a full relaxation of the atomic positions and
dynamical matrix calculations was performed.
55
3. Experimental Methods
3.1. Catalyst synthesis
3.1.4. Synthesis of magnesium oxide, MgO, catalyst
MgO catalyst was synthesized using a modified thermal decomposition method.30
In a typical synthesis, 2.23g (8.7 mmol) of Mg(NO3)26H2O (Sigma-Aldrich) were
dissolved in 40 ml methanol and then 1 ml water was added. A 30 ml methanol solution
containing 0.7 g (17.4 mmol) of NaOH was added drop-wise to the resulting solution under
reflux temperature. After 30 minutes a white precipitate was collected by centrifugation
(Thermo Sorvall™ Legend™ XT). The isolated precipitate was washed three times using a
1:1 ratio solution of ethanol/water and then separated using centrifugation. The resulting
wet samples were dried at 80 °C overnight. The resulting dry magnesium hydroxide solid
was ground using a mortar and pestle and calcined at 800 °C in a calcination oven (Thermo
Lindberg™ Blue M). Here, a ramping rate of 10 °C/min for 3 hours was used under an
oxidizing atmosphere with an air flow rate of 50 ml/min. Natural convection was used to
cool down the samples.
3.1.5. Synthesis of MgO/SiO2 catalysts
Two methods of preparation are investigated in this study, i.e. incipient wetness
impregnation (IWI) and wet-kneaded (WK). The incipient wetness impregnation was done
using final Mg/Si mass ratio of 1. Precursor used in this method is Mg(NO3)26H2O (Sigma-
Aldrich) in water, impregnated on fumed silica (Cabot). The solid is then dried overnight
under ambient condition, followed by drying at 80°C overnight, before further calcined at
800 °C. The wet-kneaded MgO/SiO2 catalysts were prepared by utilizing some of the
magnesium hydroxide material obtained in Section 3.1.1 by a thermal decomposition
56
method before its calcination. Instead of calcining, the hydroxide was wet-kneaded with
fumed silica (Cabot).13 The corresponding amounts of silica and magnesium hydroxide
were wet-kneaded in deionized water for 4 hours, centrifuged, dried overnight at 80 °C and
calcined. In Chapter 4 and 5, 800°C was chosen as calcination temperature for both
catalysts. In Chapter 4, the catalyst is labeled as WK (1:1), while in Chapter 5, the catalysts
are labeled as MgSi-WK and MgSi-IWI for comparing between WK and IWI methods,
and a WK catalyst calcined at 500°C is labeled as MgSi-WK2. In Chapter 6, for comparison
between unpromoted and promoted WK catalysts calcined at 500°C, the unpromoted
catalyst is simply referred to as MgSi.
3.1.6. Synthesis of promoted wet-kneaded MgO/SiO2 catalysts.
Following synthesis the wet-kneaded MgO/SiO2 (1:1) catalyst, drying is instead
carried out at room temperature overnight, and the catalyst was then impregnated with
transition metals, i.e., Cu or Zn. Copper nitrate trihydrate (Alfa Aesar) and zinc nitrate
hexahydrate (Sigma) were used as precursors. The catalysts were then dried at room
temperature overnight and further calcined at 500 °C for 3 hours. For comparison purpose,
an unpromoted catalyst is also calcined at 500 °C for 3 hours. These catalysts are labeled
as CuMgSi and ZnMgSi, while the corresponding binary reference catalysts, e.g. Cu(Zn)-
SiO2, Cu(Zn)-MgO, are labeled as CuSi, ZnSi, CuMg and ZnMg, respectively.
3.2. Catalytic reactivity study
The catalytic tests were performed in a Microactivity-Reference fixed bed reactor
from PID Eng Tech (Spain). A quartz tube reactor was used with the quartz wool positioned
so as to support the catalyst bed (0.1 g, pelletized, crushed and sieved to 100-150 µm
particle size). Additional SiO2 powder (Sigma) was used to increase the bed length so as
57
to maintain the plug flow condition. Ethanol was delivered with helium gas by bubbling
the gas through a chilled ethanol saturator with a total flow of 50 ml/min. The bubbler
temperature was varied to manipulate the weight hourly space velocity (WHSV). The hot
box temperature in the reactor was set at 100 °C to prevent any reactant or product
condensation. Prior to the reaction, the catalyst was activated by heating it to 500 °C at a
heating rate of 10 °C/min and then held at that temperature for 1 hour under 30 ml/min He
flow. The reactions were performed at 375 °C. The products were kept in the vapor phase
and then analyzed using a gas chromatograph equipped with an FID detector and Restek
RT-Q-Bond column. The reactant ethanol and principal products, i.e., ethylene,
acetaldehyde and 1,3-BD, were quantified based on the calibration carried out using a
standard reference mixture (Praxair).
Titration experiment was carried out to poison both basic and acidic sites. To poison
basic sites, probe molecules, i.e. CO2 and propionic acid, were used. Poisoning acidic sites
were carried out by using NH3 as the probe molecule, and by post-treatment using NaOH.
For this post-treatment method, the catalyst was impregnated with a very dilute NaOH
solution, with final catalysts containing 250, 500, and 1000 ppm NaO, and let dry at room
temperature without further thermal treatment. In a typical titration experiment, the catalyst
was let to achieve a steady-state condition at a selected WHSV and reaction temperature.
Probe molecules were then co-fed into the reactor using MFC for CO2 and 1% NH3 in N2,
while propionic acid was delivered using a chilled saturator containing mixture of
propionic acid/ethanol (3:7). After a new steady-state is achieved, the feed was reverted
back to only ethanol to check for the activity recovery.
58
3.3. Catalyst characterization
Unpromoted catalysts, i.e. MgSi-IWI and MgSi-WK, were characterized using
HS-LEIS, XRD, BET surface area measurement, and combination of in-situ IR and UV-
Vis measurements. TPRS experiments were also run using MgSi-WK. Transition metal-
promoted catalysts were characterized using XRD and BET surface area measurement,
ICP-OES, XPS, STEM, in-situ IR and UV-Vis, and also operando XAS experiments.
3.3.1. High-sensitivity low energy ion scattering (HS-LEIS)
The unpromoted IWI and WK catalysts (1:1), calcined at 800 °C, were prepared for
analysis by dispersing into an appropriate sample crucible for a heatable sample holder for
the LEIS spectrometer, ION-TOF Qtac100, and then compacting it with a sample press. The
crucible was then affixed to a sample holder with an integrated cartridge heater and a
thermocouple was placed in a hole on the crucible.
After being placed in vacuum, the temperature of a sample was raised to about 50°C
for outgassing. O2 was then introduced into the preparation temperature at an unmeasured
pressure likely between 100 and 300 mbar. The temperature of the sample was then
increased at a rate of 10°C/min to a maximum temperature of 500°C. This temperature was
held for 60 min, at which time the temperature was allowed to decrease and the preparation
chamber was evacuated. The sample was then transferred into the analysis chamber.
Charge neutralization was invoked during spectra acquisition and sputtering. For primary
ion beam, the following parameter was used: 3.0 keV He+, 1500 1500 m raster, at 2 x
1014 ions cm-2 cyc-1, 3000 eV pass energy. The following conditions were applied during
sputtering: 1.0 keV Ar+, 2000 x 2000 m raster, 5 x 1014 ions cm-2 cyc-1.
59
3.3.2. XRD and BET surface area
Bulk structural information of the catalysts was characterized using XRD. XRD
patterns were obtained using PANalytical Empyrean powder X-ray diffractometer using
Cu Kα1,2 with λ=1.5418 Å operating at 45 kV. Measurements were carried out between
2θ=10° and 100° using a step size of 0.05°. The BET specific surface areas of the catalysts
were determined by nitrogen adsorption at 77 K on a Micromeritics ASAP 2010
instrument. All samples were degassed under nitrogen flow at 623 K for 12 h before the
measurements.
3.3.3. Transition metal concentration measurements.
The weight transition metal concentration of Cu- and Zn-promoted MgO/SiO2
catalysts was determined using Inductively Coupled Plasma-Optical Emission
Spectroscopy (ICP-OES, PerkinElmer Optima 2000 DV). About 10 mg of catalyst was
digested using 40 ml solution containing 1:1:1 H2O, HCl and HNO3. Cu concentration was
measured to be 0.8%, similar to that used by Weckhuysen and coworkers18,19 while Zn was
2.5%, close to that reported by Larina et al.15
The XPS measurements were carried out to corroborate the results of ICP-OES with
a PHI 5600ci instrument using a non-monochromatized Al Kα X-ray source. The pass
energy of the analyzer was 58.7 eV, acquisition area with diameter of ~800 um and the
scan step size was 0.125 eV. Binding energies were corrected for charging by referencing
to the C 1s peak at 284.8 eV. Atomic concentrations were calculated from the areas under
individual high-resolution XPS spectra using manufacturer-provided sensitivity factors.
60
3.3.4. Scanning transmission electron microscopy
The morphology of the catalyst particles was investigated using a dedicated
Scanning Transmission Electron Microscope (STEM) (Hitachi 2700C) operating at 200
kV.
3.3.5. In-situ spectroscopy
Diffuse Reflectance Infrared Spectroscopy (DRIFTS) was used to probe the
composition and changes in hydroxyl (OH) groups on the catalyst surface under dehydrated
conditions. A Thermo Nicolet iS50 infrared spectrometer equipped with a Mercury-
Cadmium-Tellurium (MCT) liquid nitrogen cooled detector was used in combination with
a Harrick Praying Mantis™ diffuse reflection accessory equipped with ZnSe windows.
About 30 mg of <100 µm catalyst samples were loaded into the DRIFTS cell. The smaller
particle size was used to ensure a uniform catalyst bed surface for spectroscopy. Similar to
the steady state reaction testing, the catalyst activation was carried out by heating it up to
500 °C at 10 °C/min and keeping it at that temperature for 1 hour under 30 ml/min He flow.
The catalyst was then cooled down to 100 °C under 30 ml/min N2 (Praxair) flow. During
the cooling reference spectra of the catalysts were acquired at 400°C, 300°C, 200°C and
100 °C. All spectra were averaged over 96 scans at a resolution of 4 cm-1.
In-situ UV-vis DRS measurements were performed using an Agilent Technologies
Cary 5000 UV-Vis- NIR spectrophotometer equipped with a Praying Mantis TM diffuse
reflection accessory. Finely ground samples (< 100µm) of supported catalyst powders were
loaded into the environmental cell (Harrick, HVC-DR2) and then UV-vis spectra were
collected in the 200-800 nm region. An MgO reflectance standard was used as the baseline.
The experimental protocol used for DRIFTS experiments was also used in the in-situ UV-
61
VIS DRS experiments. The Kubelka-Munk function was calculated from the absorbance
of the UV-vis DRS. The edge energy (Eg) for allowed transitions was determined by
finding the intercept between the straight line and the abscissa on the Tauc plot derived
from the UV-Vis spectra.
3.3.6. Acid-base characterization using pyridine, NH3, CO2, and methanol as probe
molecules
A Thermo Nicolet iS50 infrared spectrometer equipped with a Mercury-Cadmium-
Tellurium (MCT) liquid nitrogen cooled detector was used with a Harrick Praying
Mantis™ diffuse reflection accessory and ZnSe windows to study the acidity and
basicity of the catalyst. About 30 mg of sample was pressed and loaded into the DRIFTS
cell. Catalyst activation was carried out by heating it up to 773 K at a rate of 10 K/min and
then held at that temperature for 1 hour under 30 ml/min air flow, in agreement with the
fixed bed experiment catalyst preparation procedure. The catalyst was then cooled down
to 373 K under 30 ml/min nitrogen (Praxair) flow. During the cooling, reference spectra of
the catalysts were acquired every 50 K. All spectra were averaged over 96 scans at a
resolution of 4 cm-1. Probe molecules, i.e. NH3, CO2 and pyridine, were used to characterize
the acidity and basicity of the catalyst. In general, the probe molecule is adsorbed on the
surface for 15 minutes shortly after the catalyst temperature is brought down to 373 K. This
step is followed by extensive purging using 30 ml/min N2 (Praxair) for 45 minutes. Spectra
were then continuously recorded every minute during which time the temperature was
increased to 723 K under 30 ml/min N2 flow. CO2 and NH3 (Praxair) gas cylinder is used
for delivery method, while pyridine delivery method involved bubbling N2 through the
pyridine saturator.
62
Methanol operando temperature programmed DRIFTS-MS sample preparation
was carried out in a similar manner. The product was continuously monitored using a
Cirrus 2 benchtop atmospheric pressure gas analysis system (MKS Instruments). Methanol
was used because it can test and yield products formed at the acidic, basic and redox sites.31
Briefly, after the catalyst activation step, the CH3OH was preadsorbed on the sample
surface as a saturated vapor at 4 °C using 50 ml/min helium as a carrier gas with a cell
temperature of 100 °C for 30 minutes. The catalyst was subsequently flushed with pure
helium at 30 ml/min for 1 hour. Spectra were then continuously recorded every minute,
while the temperature was ramped up to 450 °C at a rate of 10 °C/min under He flow.
Unless stated otherwise, reference spectra obtained at the corresponding temperatures were
subtracted from the acquired spectra to eliminate contribution from the catalysts.
Calibration of methanol and CO2 was performed using a mixture of both products with He
at different concentrations, while formaldehyde - by reactive calibration of methanol
dehydrogenation over Cu/SiO2 catalyst. The reaction was kept at low conversion to limit
the occurrence of secondary reactions, forming such molecules as dimethoxymethane and
methyl formate. A mass balance for the reaction system was then calculated to determine
the response factor of the formaldehyde.
3.4. Reaction mechanism study using in-situ DRIFTS spectroscopy and TPRS
A Thermo Nicolet iS50 infrared spectrometer equipped with a Mercury-Cadmium-
Tellurium (MCT) liquid nitrogen cooled detector was used with a Harrick Praying
Mantis™ diffuse reflection accessory and ZnSe windows to study the nature of the
hydroxyl groups, as well as the adsorbates on the catalyst surface. Catalyst activation is
carried out according to the procedure mentioned in Section 2.3. To monitor reactive
63
surface intermediates, ethanol was pre-adsorbed onto the sample surface as a saturated
vapor at 298 K using 30 ml/min nitrogen as a carrier gas at 373 K for 20 minutes.
Physisorbed molecules were removed with pure nitrogen at 30 ml/min for 40 minutes.
Liquid ethanol (200 proof, Koptec), crotonaldehyde (Acros organics, +99%) and crotyl
alcohol were used (Sigma, 96%). For acetaldehyde DRIFTS experiments, a gaseous
mixture of 5% acetaldehyde in nitrogen (Praxair) was used. Crotonaldehyde and crotyl
alcohol were handled with extra caution due to their toxicity. In particular, transporting the
chemical was done in the hood to an enclosed, chilled bubbler (2-4°C). The enclosed,
chilled bubbler was then installed to the gas flow delivery system while still being chilled.
Chilled bubbler lowered the partial pressure of crotonaldehyde and crotyl alcohol, further
limiting exposure to the vapor. Spectra were then continuously recorded every minute
during which time the temperature was increased to 723 K with or without the continuous
vapor flow of the reactants.
Four types of infrared spectra subtractions were applied. First, only instrumental
background was subtracted from the catalyst spectra acquired. This method was used in
Figure 4.3. Second, dehydrated catalyst spectra at 100 °C was subtracted from the spectra
of the adsorbed reactants at different temperature. This method was used in Figure 4.5.
Third, in temperature programmed desorption DRIFTS experiments, dehydrated catalyst
spectra at the exact same temperature were subtracted from the spectra of the adsorbed
reactant. This method was used in Figures 4.4-7. Fourth, in temperature programmed
desorption DRIFTS experiments, in the presence of the vapor reactant, a spectrum
containing catalyst and adsorbed surface species at 373 K was subtracted from the spectra
64
containing contributions of the catalyst, adsorbed surface species and the vapor-phase at
specific temperature. This method was used in Figures 4.9-11.
Temperature-programmed reaction spectroscopy (TPRS) was carried out using an
Altamira Instruments system (AMI-200) connected to Dymaxion Dycor mass spectrometer
(DME200MS). Approximately 30 mg of catalyst was loaded into a glass U-tube fixed-bed
reactor and held in place by quartz wool. Prior to measurement, the catalyst was first pre-
treated under 10% O2/Ar (Airgas, certified, 9.99% O2/Ar balance) at 500°C for 1 hour.
After pretreatment, the catalyst temperature was brought down to 100°C. At this
temperature, ethanol was preadsorbed for 15 minutes, followed by degassing using argon
for 45 minutes. The vapor delivery system followed that of in-situ DRIFTS study. Finally,
the fixed-bed reactor was heated at ~10°C/min to 450°C in the flowing reactant gases and
the evolution of the products was monitored with the online mass spectrometer.
Acetaldehyde was delivered using a mixture of 5% acetaldehyde in nitrogen (Praxair).
Another experiment involved preadsorbing acetaldehyde on the surface of catalyst,
followed by degassing with argon for 45 minutes, and temperature increase at 10°C/min to
450°C under constant ethanol/argon flow. The ethanol saturator is chilled at 2°C using ice
bath at all times.
3.5. Operando XANES and EXAFS spectroscopy during ethanol reaction to 1,3-
BD over Cu- and Zn-promoted MgO/SiO2 catalysts
Operando X-ray absorption spectroscopy (XAS) was performed at the beamline BL2-2 at
Stanford Synchrotron Radiation Lightsource (SSRL), SLAC National Accelerator
Laboratory. The Cu and Zn K-edge data were collected in transmission mode. For the
measurements, the sample powder was loaded into a quartz tube with 0.9 mm inner
65
diameter and 1.0 mm outer diameter which was then mounted into the Clausen plug-flow
reaction cell.54 Ethanol vapor was delivered into the system using a temperature-controlled
saturator to manipulate the space velocity. He was bubbled through the saturator and fed
into the reactor. Prior to the spectroscopy study under reaction condition, the catalyst was
pretreated at 450 °C for 1 hour under constant He flow. The operando measurements were
done at 100, 200, 300 and 400 °C under constant ethanol flow. After reactor temperature
reached 400 °C, the system was allowed to equilibrate for 2 hours and XAS spectra were
repeatedly taken. The operando condition was ensured by allowing the vapor-phase into a
dedicated RGA Mass Spectrometer (RGA, Stanford research system). Standard reference
compounds, CuO (Alfa Aesar), ZnO (Alfa Aesar), Cu2O (Alfa Aesar) and synthesized
reference materials, i.e. CuMg, ZnMg, CuSi, and ZnSi, were pressed into the pellets and
measured under ambient conditions.
66
References
(1) Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49 (20), 14251–14269.
(2) Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6 (1), 15–50.
(3) Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54 (16), 11169–11186.
(4) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 48 (17), 13115–13118.
(5) Blöchl, P. E. Phys. Rev. B 1994, 50 (24), 17953–17979.
(6) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77 (18), 3865–3868.
(7) Henkelman, G.; Jónsson, H. J. Chem. Phys. 1999, 111 (15), 7010–7022.
(8) Heyden, A.; Bell, A. T.; Keil, F. J. J. Chem. Phys. 2005, 123 (22), 224101/1-
224101/14.
(9) Fukui, K. J. Phys. Chem. 1970, 74 (23), 4161–4163.
(10) Fukui, K. Acc. Chem. Res. 1981, 14 (12), 363–368.
(11) Hratchian, H. P.; Schlegel, H. B. J. Phys. Chem. A 2001, 106 (1), 165–169.
(12) Jensen, F. Introduction to Computational Chemistry, 2nd ed.; John Wiley & Sons
Ltd., 2006.
(13) Kvisle, S.; Aguero, A.; Sneeden, R. P. A. Appl. Catal. 1988, 43 (1), 117–131.
(14) Angelici, C.; Weckhuysen, B. M.; Bruijnincx, P. C. a. ChemSusChem 2013, 6 (9),
1595–1614.
(15) Larina, O.; Kyriienko, P.; Soloviev, S. Catal. Letters 2015, 1–7.
(16) Zhang, M.; Gao, M.; Chen, J.; Yu, Y. RSC Adv. 2015, 5 (33), 25959–25966.
(17) Janssens, W.; Makshina, E. V.; Vanelderen, P.; De Clippel, F.; Houthoofd, K.;
Kerkhofs, S.; Martens, J. A.; Jacobs, P. A.; Sels, B. F. ChemSusChem 2014, 8 (6),
994–1008.
(18) Angelici, C.; Meirer, F.; van der Eerden, A. M. J.; Schaink, H. L.; Goryachev, A.;
Hofmann, J. P.; Hensen, E. J. M.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS
Catal. 2015, 5 (10), 6005–6015.
(19) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A.
ChemSusChem 2014, 7 (9), 2505–2515.
(20) Che, M.; Tench, A. J. Adv. Catal. 1982, 31, 77–133.
(21) Sushko, P. V.; Shluger, A. L.; Catlow, C. R. A. Surf. Sci. 2000, 450 (3), 153–170.
(22) Kantorovich, L. N.; Holender, J. M.; Gillan, M. J. Surf. Sci. 1995, 343 (3), 221–
239.
(23) Chizallet, C.; Petitjean, H.; Costentin, G.; Lauron-Pernot, H.; Maquet, J.;
Bonhomme, C.; Che, M. J. Catal. 2009, 268 (1), 175–179.
(24) Chizallet, C.; Bailly, M. L.; Costentin, G.; Lauron-Pernot, H.; Krafft, J. M.; Bazin,
P.; Saussey, J.; Che, M. Catal. Today 2006, 116 (2), 196–205.
(25) Di Valentin, C.; Del Vitto, A.; Pacchioni, G.; Abbet, S.; Wörz, A. S.; Judai, K.;
Heiz, U. J. Phys. Chem. B 2002, 106 (46), 11961–11969.
(26) Branda, M. M.; Rodríguez, A. H.; Belelli, P. G.; Castellani, N. J. Surf. Sci. 2009,
603 (8), 1093–1098.
(27) Bailly, M.; Chizallet, C.; Costentin, G.; Krafft, J.; Lauronpernot, H.; Che, M. J.
Catal. 2005, 235 (2), 413–422.
(28) Baltrusaitis, J.; Hatch, C.; Orlando, R. J. Phys. Chem. A 2012, 116 (30), 7950–
7958.
(29) Haas, P.; Tran, F.; Blaha, P. Phys. Rev. B 2009, 79 (8), 85104.
67
(30) Wu, C.-M.; Baltrusaitis, J.; Gillan, E. G.; Grassian, V. H. J. Phys. Chem. C 2011,
115 (20), 10164–10172.
(31) Badlani, M.; Wachs, I. E. Catal. Letters 2001, 75 (3–4), 137–149.
68
Chapter 3
Computational Study of Ethanol to 1,3-BD
Reaction Mechanisms
Abstract ........................................................................................................................... 68
1. Introduction ........................................................................................................ 69
2. Computational Results ....................................................................................... 73
2.1. Reaction Pathways .................................................................................. 73
2.1.1. Ethanol dehydrogenation and dehydration ............................................. 73
2.1.2. Aldol condensation ................................................................................. 80
2.1.3. Prins condensation .................................................................................. 85
2.1.4. 1-Ethoxyethanol formation ..................................................................... 88
2.2. Details of the free-energy profiles .......................................................... 89
2.2.1. Elimination/redox reaction of ethanol .................................................... 90
2.2.2. C-C bond formation ................................................................................ 90
2.2.3. Proton transfer ......................................................................................... 91
3. Discussion ............................................................................................................ 92
4. Conclusion .......................................................................................................... 98
References ...................................................................................................................... 100
Abstract
In this work, we performed periodic Density Functional Theory calculations and
explored reactive pathways of ethanol catalysis to catalytically form 1,3-butadiene on
undoped MgO surface. We have identified critical reactive intermediates, as well as
thermodynamic and kinetic barriers involved in the overall reactive landscape. The overall
free energy surface was explored for the highly debated reaction mechanisms, including
Toussaint’s aldol condensation mechanism, Fripiat’s Prins mechanism and mechanism
based on Ostromislensky’s hemiacetal rearrangement. Thermodynamics and kinetics data
calculated showed four rate limiting steps in the overall process. In particular, ethanol
69
dehydration to form ethylene possessed lower energy barrier than dehydrogenation to yield
acetaldehyde suggesting competing reactive pathways. C-C bond coupling to form
acetaldol (3-hydroxybutanal) is preceded with 16 kcal/mol forward reaction barrier. Direct
reaction of ethylene and acetaldehyde proceeds with a free energy barrier of 29 kcal/mol
suggesting that Prins condensation is an alternative route. Finally, thermodynamic stability
of 1-ethoxyethanol prevents further reaction via hemiacetal rearrangement. The results
here provide a first glimpse into the overall 1,3-butadiene formation mechanism on
undoped MgO reactive sites in light of the vast literature discussing variety of the proposed
mechanistic pathways mostly based on conventional homogenous organic chemistry
reactions.
1. Introduction
Since the rapid expansion of coal industry in the 18th century World has relied on
non-renewable sources for organic chemicals 1. Currently, petroleum and natural gas are
the main feedstocks of relatively inexpensive carbon source 2. Biomass can serve as a
sustainable and renewable carbon source to generate chemicals and bio derived ethanol
catalytic upgrading has been proposed as a viable route for biomass valorization 3,4. In
particular, ethylene, propylene, ethyl acetate, n-butanol and isobutene are some of the high
value chemicals that can be derived from ethanol 5–14. Furthermore, 1,3-butadiene, the most
important monomer for synthetic rubber, has been produced via catalytic processing of
ethanol during World War II in USSR and USA, using Lebedev and Ostromislensky
processes, respectively 15. The former utilized catalytic conversion of ethanol to 1,3-
70
butadiene via one-step on MgO/SiO2 catalysts 16, while the latter utilized a two-step process
with the first step ethanol dehydrogenation to acetaldehyde over Cu/SiO2 catalysts 3,17
followed by acetaldehyde and ethanol coupling to 1,3-butadiene over a tantalum-based
catalyst. Catalysts for the one-step process were reported to have achieved ~50-60% yield
18,19, while the two-step process could attain over 60% yield, with purity of about 98-99%
at 300-350°C. Recent abundance of shale gas resulted in a different catalytic cracker
product distribution dominated by ethylene 3. This caused a worldwide shortage of C4
hydrocarbons, such as 1,3-butadiene. Since ethanol can be produced using variety of
biomass sources including fermentation and gasification, it recently emerged as the green
route to catalytically form 1,3-butadiene 3,4.
The biggest obstacle in ethanol catalysis to form 1,3-butadiene is relatively low
selectivity and the resulting yields of the desired product. Angelici et al. reported 74%
conversion with 49% selectivity on CuO/MgO-SiO2, while Makshina et al. described a
similar catalyst that attained 97.5% conversion with 58.2% selectivity 18,20. Most recently,
a CuO/HfO/ZnO catalyst was reported to have achieved 99% conversion with 71.1%
selectivity, e.g. ~70% yield 19. In general, doped-MgO supported on silica 18,20–29
(Lebedev’s catalyst) or mixed (supported) oxides 16,19,30–36 were used. Various transition
metal dopants (Zn, Cu, and Ag) were used to improve MgO/SiO2 catalyst performance, as
well as different synthesis methods and composition of MgO/SiO2 were investigated, such
as Kvisle’s wet kneading method 26 and the utilization of clay and sepiolite as the support
21,28. However, multitude of byproducts, including ethylene, C4 oxygenates and olefins,
diethyl ether, acetaldehyde and even acetone are still detected implying high separation.
This lack of the kinetic control over the ethanol-to-1,3-butadiene catalytic process and poor
71
understanding of the fundamental mechanistic steps involved have hindered the
development of catalysts with reasonable performance. The generally accepted one-step
catalytic mechanism involves dehydrogenation of ethanol to acetaldehyde which then
undergoes C-C coupling via aldol condensation mechanism to yield crotonaldehyde.
Crotonaldehyde is further hydrogenated via MPV (Meerwein-Ponndorf–Verley) reduction
with ethanol and the resulting crotyl alcohol is dehydrated give butadiene37,38 as shown in
Figure 3.1a. In addition, Fripiat and Ostromislensky proposed two other possible reaction
pathways 39,40. Fripiat suggested Prins-like mechanism involving both dehydration and
dehydrogenation reactions producing ethylene and acetaldehyde, as shown in Figure 3.1b.
The C=O group is hydroxylated in the presence of Brønsted acid and reacts with ethylene
opening the double bond. The resulting 3-buten-2-ol is then dehydrated to yield 1,3-
butadiene 40. Ostromislensky’s version of the reaction mechanism shown in Figure 3.1c
involves the hemiacetal rearrangement between ethanol and acetaldehyde to yield 1-
ethoxyethanol that later converts to butane-1,3-diol 3. Two computational studies by
Chieregato et al. 41 and Zhang et al. 42 attempted to unravel the overall reaction mechanism.
Zhang et al. performed calculations using density functional generalized gradient
approximation (GGA) and focused on the very first step of the mechanism, e.g.
dehydrogenation of the alcohol. Stepped MgO surface was predicted to have lower energy
barrier than flat surface for this reaction 42. Chieregato et al., on the other hand, proposed
an entirely different mechanism using cluster type calculations and Gaussian basis set.
They ruled out crotonaldehyde and crotyl alcohol as possible intermediates and concluded
that acetaldehyde would react with a carbanion resulting from ethanol C-H cleavage. In
this work we performed ethanol catalytic coupling to 1,3-butadiene using a kink Mg atom
72
at a step-edge MgO (100) as a model catalyst surface in accordance with the recent works
that suggest MgO as a bifunctional catalyst 43–53. The energetics and structure of key
reactive intermediates, e.g. acetaldehyde, crotonaldehyde, and crotyl alcohol, for
Lebedev’s reaction 16 based on the proposed mechanism by Toussaint et al. 38,54, as well as
the proposed Fripiat’s Prins and Ostromislensky mechanistic pathways were explored to
determine the kinetic limitations of ethanol catalytic coupling to 1,3-butadiene on MgO.
Figure 3.1. Reaction mechanisms proposed for ethanol to 1,3-butadiene; (a) Toussaint’s
generally accepted mechanism, (b) Fripiat’s Prins mechanism, (c) Ostromislensky’s
hemiacetal rearrangement.
73
2. Computational results
2.1. Reaction Pathways
2.1.1. Ethanol Dehydrogenation and Dehydration
We begin with the first step of ethanol catalytic transformation into acetaldehyde
or into ethylene for which the computed free-energy profiles are shown in Figure 3.3.
States 1A-1C and 5A-5C in Figure 3.2 demonstrate the reaction pathways for ethanol
dehydrogenation and dehydration to form acetaldehyde and ethylene, respectively. The
corresponding TSs are labelled as 1B(TS) and 5B(TS). In both cases the reaction starts
from the structures formed upon spontaneous dissociative adsorption of ethanol whereby
proton is abstracted either by the edge or terrace oxygen atoms. The calculated relative free
energies for the configurations 1A and 5A are -13.5 and -10.5 kcal/mol, respectively, at
450 oC. In 1B TS, the surface bound proton becomes coordinated to the proton leaving -
carbon atom. The H….H distance is 0.84 Å, which is 0.07 Å longer than the equilibrium
H-H bond distance in hydrogen molecule in 1C. Compared to 1A, the distance between the
hydrogen atom and lattice O4C has significantly increased from 0.98 Å to 1.61 Å.
Furthermore, acetaldehyde is coordinated to Mg3c via oxygen atom while also accepting
some electron density from in the terrace O5c resulting in a distorted chemisorbed structure.
During the dehydration step hydrogen atom is adsorbed on terrace oxygen atom as shown
in Fig. 3.2 (5A). During the reaction over transition state 5B one of the β-hydrogen atoms
becomes oriented towards O4C with C-H and O4C-H of 1.45 and 1.22 Å, respectively. The
final state 5C results in two hydroxyl groups formed on the MgO surface with ethylene
molecule loosely coordinated to the surface. Relative electronic energies at T=0 K and the
corresponding free energy values at 450 oC for the reaction profile shown in Figure 3.3 are
74
provided in Table 3.1. In particular, at 450°C the forward free-energy barrier for the
reaction 1A 1C is 39.6 kcal/mol while the barrier for the reverse barrier is only 20.5
kcal/mol suggesting that the acetaldehyde formation is endergonic. Ethylene formation via
5A 5C has barrier 33.5 kcal/mol but the reverse barrier is higher with 38.3 kcal/mol.
While both acetaldehyde and ethylene are typically observed as reaction byproducts of
ethanol catalytic coupling 18,20,22–25, the stability of ethylene vs acetaldehyde is intriguing
but not surprising. Ethanol is known to undergo intramolecular dehydration in the presence
of acidic and basic surface sites 55, while acetaldehyde formation in general needs redox
metals 56. Hence for undoped MgO catalyst ethylene generation is expected and preferred
over acetaldehyde.
The reaction pathways of ethanol dehydration and dehydrogenation products
further proceed via two main reaction mechanisms discussed in detail in Sec.3.1.2 and
3.1.3: aldol condensation and Prins condensation reaction. The aldol condensation pathway
entails acetaldehyde transformation into its enolate form followed by the reaction with the
molecular acetaldehyde to form a C-C bond. The resulting C4 intermediate then undergoes
several steps of intermolecular proton transfer with the surface and ethanol to yield 1,3-
butadiene. Prins condensation entails C-C bond formation via reaction of acetaldehyde
and ethylene followed by proton transfer steps. In these mechanisms, the proton diffusion
through the surface leads to water release from the surface.
75
Table 3.1 Electronic and free energy values of the stationary points calculated at 0 K and
723 K, respectively.
State
Electronic
energy
(kcal/mol)
Referenced
free energy
(kcal/mol)
State
Electronic
energy
(kcal/mol)
Referenced
free energy
(kcal/mol)
1A -40.6 -13.5 4F -45.3 -8.1
1B 3.9 26.2 4G -43.2 -6.1
1C -16.6 5.6 4H -52.0 -46.2
2A -14.3 -8.5 4I -37.8 -37.7
2B 4.1 7.4 4J -5.5 -5.3
2C -22.2 -20.3 4K -18.7 -24.0
2D -17.7 -17.3 5A -39.9 -10.5
2E -26.2 -28.8 5B -2.6 23.0
2F -22.5 -23.7 5C -32.0 -15.3
2G -15.9 -7.6 6A 11.6 -0.2
2H -19.4 -10.8 6B 25.6 28.6
2I -12.5 -1.6 6C -34.8 -22.9
2J -11.4 -1.6 6D 3.7 15.8
2K -6.2 4.2 6E -26.8 -18.9
2L -15.4 -8.2 6F 9.5 15.3
2M 9.3 15.7 6G -6.0 -0.7
2N 2.2 3.2 6H -4.4 -2.9
2O 6.8 -18.2 6E i -21.8 -16.0
3A -24.1 -18.8 6E ii -30.5 -27.2
3B -14.6 -6.8 6E iii -32.5 -27.3
3C -19.8 -15.2 6E iii 1 2.2 5.6
3D -27.3 -27.8 6E iii 2 -13.8 -9.6
3E -1.3 -0.2 6E iii 3 -11.5 -12.7
3F -15.4 -15.4 6E iv 43.9 44.3
3G -15.6 -20.4 6E v -21.4 -25.4
4A -53.1 -14.6 7A -51.2 -26.4
4B -52.5 -4.8 7B -30.6 2.5
4C -64.4 -22.1 7C -33.4 3.9
4D -68.3 -32.0 7D -30.5 6.4
4E -42.5 -5.8 7E -44.8 -8.7
76
1A 1B (TS) 5A 5B (TS)
1C 5C
2A 2B (TS) 2C 2D (TS)
2E 2F 2G (TS) 2H
2I 2J 2K (TS) 2L
77
2M (TS) 2N 2O
3A 3B (TS) 3C 3D
3E (TS) 3F 3G 4A
4B (TS) 4C 4D 4E (TS)
4F 4G (TS) 4H 4I
78
4J (TS) 4K
6A 6B (TS) 6C 6D (TS)
6E 6F (TS) 6G 6H
6E i (TS) 6E ii 6E iii 6E iv (TS)
6E v 6E iii 1 (TS) 6E iii 2 6E iii 3
79
7A 7B (TS) 7C 7D (TS)
7E
Figure 3.3 Free-energy profiles for (a) ethanol dehydrogenation to form acetaldehyde
and (b) ethanol dehydration to ethylene.
(a) (b)
Figure 3.2 All stable intermediates and transition states calculated following the reaction
pathways. (1A-1C): ethanol dehydrogenation to acetaldehyde; (2A-2O): acetaldehyde
aldol condensation to 3-hydroxybutanal (acetaldol) followed by proton transfer to
crotonaldehyde; (3A-3G): MPV (Meerwein–Ponndorf–Verley) reduction of
crotonaldehyde to 1,3-butadiene; (4A-4K): acetaldol MPV reduction to butadiene; (5A-
5C): ethanol dehydration to ethylene; (6A-6E iii 3): Prins condensation of acetaldehyde
and ethylene; (7A-7E): ethanol and acetaldehyde nucleophilic addition reaction
(Ostromislensky’s hemiacetal rearrangement).
80
2.1.2. Aldol condensation
Classical aldol condensation mechanism requires one of the acetaldehyde
molecules to be in its enolate state.57 In this work the enolate state 2C (see Fig. 3.2) was
obtained via proton transfer of β-hydrogen to terrace atom O5C yielding a hydroxyl group
via low energy barrier 2B TS from the initial stable strongly adsorbed acetaldehyde
molecule in 2A with free-energy of -8.5 kcal/mol relative to the reference state. The
forward barrier for the reaction step 2A 2C is 16 kcal/mol as shown in Figure 3.4. This
mechanism is facilitated by the C=O bond elongated from 1.21 Å58 to 1.43 Å due to strong
interaction with the surface oxygen atoms. In TS configuration, one of hydrogen atoms
from the methyl group establishes a hydrogen bond with surface oxygen atom causing an
elongation of the corresponding C-H bond from 1.10 to 1.29 Å. State 2C represents a stable
configuration with sp2 hybridized carbon enolate atoms and surface hydroxyl group.
For the aldol condensation to take place, the hydrogen atom bound to the surface
needs to be in a close proximity to the enolate molecule requiring it to diffuse to the edge
O4C atom. This transition proceeds via transition state structure 2D with the forward barrier
of only 3 kcal/mol. The next step is the physisorption of a second acetaldehyde molecule
on the surface in 2F, preceding the C-C bond formation via aldol condensation (2F-2H).
The TS for this step (the structure 2G) shows the coordination between enolate and
acetaldehyde with the reactive β-carbon of enolate and the -carbon of acetaldehyde
adsorbed on surface site Mg3C establish a C-C bond of length of 2.07 Å. The forward IRC
analysis for this reaction shows the formation of C-C bond between the two reactive carbon
atoms. The length of the latter bond in the stable structure 2H is 1.64 Å. Formation of the
81
acetaldol (3-hydroxybutanal) in 2H is preceded with forward reaction barrier of 16.1
kcal/mol.
Once 3-hydroxybutanal is formed, the aldol needs to lose a hydrogen atom to the
surface to undergo dehydration to yield crotonaldehyde. For this step to take place, a
reactive O4C is required. Assuming transient proton diffusion between O4C and O5C atoms,
proton abstraction takes place followed by the transformation of the trans-isomer, as
depicted in 2I, into 2J via aergonic steps. The activation energy from cis- to trans-
crotonaldehyde reported is ~13 kcal/mol 59. The molecule subsequently loses hydrogen
Figure 3.4 Free-energy profiles for aldol condensation
pathway.
82
atoms to the MgO surface via step 2J2L (ΔGA,forward = 5.9 kcal/mol) and desorbs from
the surface after breaking the C-O bond via sequence 2L2M(TS)2O (ΔGA,forward = 24
kcal/mol) yielding the structure 2O with a newly formed surface site O3c which under
reactive conditions can recombine with protons to form H2O reforming the original Mg3c
site.
Formation and desorption of crotonaldehyde in 2N-2O agrees very well with the
occasional gas phase byproduct observations 22,37. Formation of the O3C surface site in 2O
is followed by the water molecule formation, which is another product of ethanol coupling
reactions. The next step in the overall mechanism is the MPV reduction of crotonaldehyde
by ethanol (3A3C) to form adsorbed acetaldehyde in 3C followed by its desorption and
proton transfer to the surface (3C3G) to form 1,3-butadiene. The corresponding free-
energy profile (2H 2O 3A 3G) is shown in Figure 3.5. The highest barrier that
we determined within this sequence was that of proton transfer to the surface in 3E TS with
27.6 kcal/mol. Reduction of the unsaturated aldehyde by hydrogen was assumed not to take
place as confirmed by the measured hydrogen content in the reaction products 30 and due
to the gas-phase thermodynamic calculation which favors the reduction by ethanol 3.
Dissociation of hydrogen on the defected MgO surface itself is a non-spontaneous process
with a relatively low activation barrier of 2.8 kcal/mol 60 whereas ethanol dissociation on
Mg3c is spontaneous as shown in Figure 3.2. Additionally, the heterolytic dissociation can
only be stabilized on a high density of 3-coordinated sites which suggests small amount of
surface hydrogen 61.
83
84
Alternatively, the 1,3-butadiene formation mechanism proposed earlier by
Ostromislensky (also vide infra) involves the hemiacetal rearrangement 39. It was argued
that ethanol can react with acetaldehyde to form 1-ethoxyethanol which will then undergo
rearrangement to butane-1,3-diol and further dehydrate to 1,3-butadiene. However, this
mechanism has been rebuffed by Quattlebaum et al. 37. The formed C-O-C bond, if it is to
be rearranged to make C-C bond, would lead to its dissociation. The identified butane-1,3-
diol, however, could be formed when acetaldol is reduced by ethanol, as shown in Figure
3.5 4A-4C. State 4C is effectively dissociated (adsorbed) butane-1,3-diol which is formed
via series of exergonic steps 2J-4A-4B TS-4C with a very low forward free-energy barrier
of 9.8 kcal/mol. The resulting adsorbed butane-1,3-diol can further undergo several steps
of proton transfer to yield 1,3-butadiene (4C-4K). In this situation, there are three
competing processes with transition states 4E, 4G, and 4J. The reaction channel with TS
4G breaks a C-O bond of the adsorbed butane-1,3-diol. The reaction 4F4H is extremely
Figure 3.5 Free-energy profiles for the MPV reduction of the resulting molecule from
aldol condensation. Red pathway indicates subsequent proton transfer of acetaldol
followed by MPV reduction of the crotonaldehyde; Blue pathway shows the direct MPV
reduction of the resulting acetaldol.
85
exergonic with very low free-energy of activation (~2 kcal/mol). It is preceded by the 4E
TS and the free-energy of activation of 26.2 kcal/mol, which is a typical value to that of
sp3 proton transfer to the reactive surface O4c atoms. The last step of this condensation
mechanism is the simultaneous C-O bond breaking and proton transfer to the surface
(4I4K). Fig. 3.2 (4J) depicts a transition state where the 1,3-butadiene is desorbing from
the surface. Interestingly, the two different MPV reduction steps yield two different
conformations of 1,3-butadiene. MPV reduction of crotonaldehyde gives s-trans
conformation, while that of acetaldol results in s-cis conformation (structures 3G and 4K,
respectively, see Fig. 3.2). The stable conformation is, however, trans 1,3-butadiene, which
makes an additional step for acetaldol reduction necessary. This last step will be cis/trans
isomerization to trans 1,3-butadiene with rather low free-energy of activation of only ~4
kcal/mol 62.
2.1.3. Prins condensation
Our data shown in Figure 3.2 suggest that on undoped MgO the ethylene formation
from ethanol will compete with that of acetaldehyde. Prins condensation is among the early
proposed mechanisms for ethanol reaction to 1,3-butadiene.40 The explicit ethanol reaction
mechanism on MgO via Prins mechanism is studied in this work and the corresponding
results are presented in this section. The corresponding structures are shown in Figure 3.2
(6A-6E) whereas the free-energy reaction profiles are displayed in Figure 3.6. The Prins
condensation pathway is the formation of C-C bond by opening the double C=C and C=O
bonds of both ethylene and acetaldehyde, respectively (6A-6C). 6B TS represents the
double bond opening of C=O with the oxygen coordinated between two surface Mg atoms
(Mg4C and Mg3C). This charge transfer to the MgO surface makes the -carbon susceptible
86
to attack by the sp2 carbon molecule. The intermolecular C-C distance is now 1.90 Å, while
that of the aldehyde C-O bond is elongated by 0.1 Å. The double bond opening results in
a C4 structure bound to the surface, as shown in Fig. 3.2 (6C). The corresponding forward
free-energy barrier is 28.8 kcal/mol. This C-C coupling step is then followed by the proton
transfer to the surface atom O5c and the simultaneous C-Olattice bond breaking (see Fig. 3.2
(6C-6E)) followed by another proton transfer from the terminal sp3 carbon to the O4c
surface atom (see Fig. 3.2 (6E-6G)). The free-energy barriers for these steps are of 38.7
and 34.1 kcal/mol, which are typical values for proton transfer reactions considered in this
work. The structure 6H represents the desorbed structure of 1,3-butadiene.
An alternative pathway for transformation of the structure 6E is that via transition
state 6E i leading to the product 6E ii (see Fig. 3.2). Here instead of the subsequent proton
transfer from the terminal carbon (as in the reaction 6E6G) the surface proton diffuses
from planar surface atom O5c to a nearby edge atom O4c via low forward barrier of 2.9
kcal/mol. The terminal proton transfer then takes place, as depicted in Fig. 3.2 (6E iii, 6E
iii 1, 6iii 2) and the computed free-energy of activation (32.9 kcal/mol) is comparable to
that for steps alternative pathway (6C6E and 6E6G). Interestingly, a cyclic TS can
also be established in another variant of this mechanism, see Fig. 3.2 (6E iv). This
mechanism yields a physisorbed molecule as a product, which is similar to methylethyl
ketone (MEK) shown in Fig. 3.2 (6E v). However, this step has rather large forward
activation energy of 71.6 kcal/mol.
87
Figure 3.6 Free-energy profiles for the Prins condensation between ethylene and
acetaldehyde. Red pathway indicates a typical route of Prins condensation; Blue
pathway shows an additional proton diffusion step in between the reaction steps; Black
pathway shows the unlikely formation of MEK.
88
2.1.4. 1-ethoxyethanol formation
Final major reaction mechanism considered in this study is the Ostromislensky’s
hemiacetal rearrangement which will be discussed in this section. The very first step in this
case is the reaction of ethanol and acetaldehyde to yield 1-ethoxyethanol, which was further
postulated to undergo a molecular rearrangement to form butane-1,3-diol. The very first
step was investigated and it was found to proceed via stationary structures 7A to 7E shown
in Figure 3.2. The free-energy profile for these steps is shown in Figure 3.7. The initial
structure 7A contains an ethoxy species formed during the chemisorption of ethanol, as
well as a molecule of acetaldehyde physisorbed to the MgO surface. The C-O bond
formation to yield 1-ethoxyethanol (via transition state 7B) has a forward barrier of 28.9
kcal/mol with several nearly isoexergonic molecular rearrangements followed by the
stabilized 7E structure with former aldehyde C-O that is still coordinated to the surface
atom Mg3c. It is apparent from Figure 3.7 that reverse barrier for the 1-ethoxyethanol
formation is almost zero. In this case 1-ethoxyethanol can behave as a thermodynamical
sink that would form in a transient fashion before reacting via other discussed pathways to
form 1,3-butadiene. Surprisingly, the free energy computed for the structure 7C, which is
a minimum on potential energy surface (PES), is slightly higher than that for the structure
7B – TS that is a first-order saddle point on PES. This unexpected result is clearly due to a
failure of the harmonic approximation used in this work to determine free-energies. This
level of theory implies that the positions of stationary points on PES and on the free-energy
surfaces are identical which is generally not true (see Ref.63 for discussion of the limitations
of harmonic transition state theory). Furthermore, this level of theory is unsuitable to
describe soft degrees of freedom such as hindered molecular rotations or long-wave lattice
89
vibrations which contribute to harmonic free-energies more than the hard ones. In our case,
the free energy for the first-order saddle point structure 7B is 28.9 kcal/mol higher than
that for the minimum 7A and, importantly, the reaction coordinate for the whole sequence
7A→7E consists of hindered rotations of the CH3CH2O- and CH3CHO- groups with the
imaginary frequencies for the TS structures 7B and 7D that are smaller than 100 cm-1. We
note, however, that the free-energies for the sequence of structures between two stable
configurations 7A and 7E (i.e. the structures 7B, 7C and 7D) are all within 4 kcal/mol and
this number is relatively small compared to the free-energy difference with respect to the
stable structure 7A. As the configuration 7D is the one with the highest free-energy on the
sequence of steps 7A→7E, we consider the difference G(7D)-G(7A)=32. 8 kcal/mol as the
effective free-energy barrier for the whole process 7A→7E.
2.2. Details of the Free-energy profiles
Three particular steps will be discussed here, namely elimination/redox reactions of
ethanol, C-C bond formation, and proton transfer.
Figure 3.7 Free-energy profile for ethanol and acetaldehyde nucleophilic addition
reaction.
90
2.2.1. Elimination/redox reaction of ethanol
Elimination reaction takes place when a substituent leaves the molecule, e.g. water leaving
ethanol, while redox reaction is defined as a reaction where a molecule loses or gains
hydrogen. 64 Dehydrogenation reaction of ethanol (redox) is the first and foremost reaction
step in all mechanisms proposed for the 1,3-butadiene formation. This oxidation step yields
hydrogen as a byproduct while also transforming ethanol into acetaldehyde, a more reactive
intermediate. The transformation 1A1C shows a rather high free-energy barrier, 39.6
kcal/mol, while the reaction itself is endergonic in nature, with ΔGRx=20.5 kcal/mol. On
the other hand, ethanol dehydration to ethylene has slightly lower activation barrier, 33.5
kcal/mol, and it is slightly exergonic with ΔGRx=-4.9 kcal/mol. Comparison of both
reactions shows that ethanol is more likely to lose water than hydrogen on undoped MgO
catalyst surface, which means that ethylene should be produced in higher amounts than
acetaldehyde. This is in agreement with the experimental reports where small amounts of
Ag, Cu or Zn are typically incorporated into the lattice structure to enhance acetaldehyde
formation 23,65–71.
2.2.2. C-C bond formation
Two C-C bond formation pathways are presented in this study. Namely, aldol condensation
and Prins condensation. The pathway 2F2H possesses a favorable activation energy,
lower than that of Prins, 16.1 v 28.8 kcal/mol. However, this reaction is thermodynamically
limited, as shown by the endergonic nature of the reaction, with ΔGRx=12.9 kcal/mol. This
suggests that aldol condensation step is kinetically favored on undoped MgO samples, with
arguably one of the most favorable steps in the whole reaction landscape, while the
exergonic nature of the Prins mechanism makes it thermodynamically favored. The
91
activation energy of the latter is also similar to the energetic barrier to other steps,
comparable to that of ethylene formation, and even lower than the activation energy of
ethanol dehydrogenation. However, the overall picture is more complex since ethylene
formed is only physisorbed onto the surface and adsorption of both reactants, i.e. ethylene
and acetaldehyde, is only -0.2 kcal/mol lower in free-energy than the reference state
suggesting that both molecules can desorb as byproducts. In accord with implications of
these results, both ethylene and acetaldehyde have been seen as byproducts of ethanol
catalytic coupling to form 1,3-butadiene.18,20,22–25
2.2.3. Proton transfer
Proton transfer steps can be further subdivided into three categories. The proton transfer
steps of the first category are those that take place between the organic molecule and the
surface, e.g. 2J2L, 3D3F, 4D4F, 4I4K, 6C6G. The second type of proton
transfer reactions is the MPV reduction taking place between two organic molecules, e.g.
3A3C and 4A4C. The last type is the proton diffusion from one site of the MgO
surface to another, e.g. 6E6E i6E ii, 2C2E, and also the understated proton diffusion
steps between 2H and 2I.
The first type of proton transfer typically exhibits moderate activation energy. Most of the
cases have ΔGA,forward = ~30 kcal/mol with only one case, i.e. 2J2L, possessing very low
activation energy of ~6 kcal/mol, possibly due to the very saturated nature of the organic
C4 compound. In the case of the MPV reduction, hydrogen atom moves from an alcohol
-carbon to open up the C=O bond of a molecule. This reduction reaction does not have a
typical activation energy but rather it depends on the nature of the C=O containing
molecule itself. The values computed for crotonaldehyde and acetaldol are ~12 kcal/mol
92
and ~10 kcal/mol, respectively. Finally, the last type of the proton transfer reaction is
proton diffusion from one MgO surface site to another. This reaction has typically very
low activation energy of ~3 kcal/mol which suggests that the water formation and
desorption are easily facilitated by the MgO catalyst.
3. Discussion
Adsorption of ethanol on both perfect and defect sites of MgO surface had been
studied previously using cluster calculation 47. It was shown that ethanol dissociated on
defect sites but not on the perfect surface. Moreover, the adsorption energy decreased with
the coordination number of the adsorption site 47. This finding also aligns well with our
calculations which show two modes of ethanol dissociation on the defect sites, i.e. 1A and
5A. In the state 1A, adsorbed molecules are coordinated on Mg3C (corner) and O4C (edge),
while in the configuration 5A the ethanol molecule is chemisorbed on Mg3C and O5C
(terrace). The energy of the former is lower than the latter (both electronic energy and
Gibbs’ free energy), indicating the difference in stability of both states. The lower
coordination O4C is very reactive and hence the spontaneous chemisorption of ethanol 45,52.
Figure 3.2 structure 5A, however, depicts the adsorption of ethanol on the same Mg atoms
(Mg3C and Mg5C), but proton adsorbing on O5C. The highly-coordinated O5C does not
possess similar deprotonation ability to its lower-coordinated counterparts, as it is already
stabilized by coordination with the neighboring atoms 45,50. This situation causes the ethoxy
oxygen to interact less strongly with the surface in Figure 5A, resulting in a relatively more
unstable state compared to Figure 3.2 structure 1A. Similarly, two new hydroxyl groups
93
are formed during the dehydration process and only weakly bound acetaldehyde and
molecular hydrogen during the dehydrogenation process in Figure 3.2 structures 5C and
1C, respectively, with the former being more stable.
A study by Zhang et al. showed a peculiar finding where ethanol dissociated on
both perfect and defected sites with energy barriers of 1.63, 1.42, and 1.30 eV, for terrace,
kink, and edge, respectively 42. Surprisingly, the molecule needed to surpass higher barrier
for kink which consisted of two low coordinated ions (Mg3C-O3C), than for stepped Mg4C-
O4C. This is in contrast to findings reported in this work and those of Branda where strong
dissociation on the defect sites takes place without any barrier 47. Finally, Chieregato et al.
41 showed that ethanol dehydrogenation over corner site of the MgO surface had an
energetic barrier of 44.7 kcal/mol on Mg3C site, as determined using cluster B3LYP/6-
31++G(d,p) DFT calculations. Furthermore, the reaction was also postulated to be slightly
exergonic with respect to the gas phase reference components, with ΔE of -1.4 kcal/mol.
Our periodic calculations, on the other hand, predict that ethanol to acetaldehyde has a
rather high energetic barrier, and it essentially represents rate-limiting step in the overall
mechanism. The free-energy barrier, based on our calculation was 39.6 kcal/mol at 450°C.
The ΔGRx for this reaction is also calculated to be +19.1 kcal/mol, which is highly
endergonic.
The free-energy values for the profiles presented in Figure 3.3-7 are listed in Table
3.2, along with the computed reaction rate constants. Based on the results presented in
Figure 3.3, ethylene is more likely to be produced than acetaldehyde due to the lower
activation energy and the exergonic nature of the reaction. As depicted in Figure 3.4 only
one TS structure has barrier higher than the desorption energy of the molecules in the
94
reference state, namely that for enolate formation (2A-2C). In the subsequent step C-C
bond coupling takes place between the enolate and the physisorbed acetaldehyde (2F-2H).
This transition state (2G) is facilitated by enol, acetaldehyde as well as the resulting C-C
product bonded by low-coordinated Mg atoms.
The resulting acetaldol, after several steps of proton diffusion and isomerization to
cis conformer, can either lose proton to the surface or undergo MPV reduction, as shown
in Figure 3.5 for steps 2J2L and 2J4C, respectively. The two different pathways show
that MPV reduction of acetaldol can be more favorable with a sharp decrease in its energy
when ethanol is adsorbed, i.e. ethanol adsorption is much more favorable than a proton
transfer from acetaldol to the surface. One should note that the overall MPV reduction
pathway of acetaldol is below the reference state, which means that all the reaction steps
are more favorable than the desorption of any adsorbates. State 4D, which results from the
subsequent acetaldehyde desorption from state 4E, is essentially an adsorbed butane-1,3-
diol. This is the basis of Ostromislensky’s reaction mechanism supported by our
calculations, although the rearrangement step from 1-ethoxyethanol to this diol has been
previously rejected.39 As mentioned before, MPV reduction of acetaldol pathway would
require an additional step to convert the cis-1,3-butadiene to trans-1,3-butadiene (ΔGRx =
~4kcal/mol), which is the more stable molecule.
The adsorption of crotonaldehyde on the defected surface shows that the C=O bond
is now lengthened from 1.25 Å (gas-phase) to 1.27 Å. This bond lengthening, also noticed
by Boronat, et al., is attributed to the back-donation of the surface antibonding orbital
π*(CO) of crotonaldehyde 72. These authors calculated three main pathways for MPV
reduction of cycohexanone by 2-butanol over a tin-zeolite catalyst and reported the most
95
favorable pathway (ΔGA,forward = ~15 kcal/mol) proceeding via formation of alkoxy species
on the surface, although their calculation was carried out on a single metal center model 72.
Furthermore, both direct MVP and H-transfer facilitated by metal hydride formation have
been reported with the former taking place over alkali-catalysts while the latter over
transition metal catalysts 73,74. A direct MPV reduction mechanism was also reported to
take place during the 5-HMF reduction by methanol on Mg3c site of MgO cluster model,
as reported by Pasini, et al. 73 with electronic energy of 27.5 kcal/mol.
Table 3.2 Computed forward and reverse reaction barriers and the corresponding reaction
rate constants.
Reaction ΔGA (kcal/mol) K (s-1)
Forward Reverse Forward Reverse
1A 1C 39.6 20.5 15.9 9.37 106
2A 2C 16.0 27.7 2.26 108 6.26 104
2C 2E 3.0 10.5 1.84 1012 9.93 109
2F 2H 16.1 3.2 2.04 108 1.60 1012
2J 2L 5.9 12.5 2.52 1011 2.59 109
2L 2N 24.0 12.5 8.66 105 2.44 109
3A 3C 12.1 8.5 3.35 109 4.17 1010
3D 3F 27.6 15.3 6.88 104 3.67 108
4A 4C 9.8 17.3 1.69 1010 8.74 107
4D 4F 26.2 2.4 1.77 105 2.86 1012
4F 4H 2.1 40.2 3.57 1012 11.0
4I 4K 32.5 18.8 2.26 103 3.18 107
5A 5C 33.5 38.3 1.15 103 38.9
6A 6C 28.8 51.5 3.07 104 4.11 10-3
6C 6E 38.7 34.7 29.6 4.93 102
6E 6G 34.1 2.3 7.45 102 3.14 1012
6E 6E ii 2.9 11.3 2.00 1012 6.00 109
6E iii 6E v 71.6 69.7 3.43 10-9 1.26 10-8
6E iii 6E iii2 32.9 15.2 1.76 103 3.92 108
7A 7E 32.8 15.1 1.86 103 4.24 108
96
Free-energy profile for the Prins pathway is depicted in Figure 3.6. Both
acetaldehyde and ethylene (6A) have rather low adsorption energies, i.e. the two C2 species
can easily desorb from the surface before going to the anticipated TS (6B). Notably, all
transition states in this mechanism have positive relative energy with respect to the
reference state. The blue pathway indicates a Prins mechanism that includes proton
diffusion which results in a slightly lower free energy of the last transition state (6E iii 1)
compared to the original red pathway, of which last transition state has a higher free energy
(6F). The final state in both variants of Prins mechanism is trans-1,3-butadiene detached
from the surface. Another step considered within the discussion of Prins mechanism is the
formation of highly energetically unfavorable cyclic TS (6E iv). Not only does it have a
large activation barrier, but it also goes to another minimum (6E v) which has a slightly
higher relative energy than the initial state (6E iii).
The Prins mechanism was originally suggested by Gruver et al. 40. The authors used
aluminated sepiolites (both ammonium-exchanged and silver-exchanged) for the butadiene
production from ethanol. On the silver exchanged catalyst, the production of ethylene and
1,3-butadiene increased exponentially with increasing contact time, while acetaldehyde
production was linear 40. The adoption of the same mechanism for MgO can be attributed
to the fact that both catalysts possess almost exclusively Lewis acid sites 75 as Prins is a
mechanism that mostly takes place on Lewis acid sites 76. The reason that this mechanism
has not been considered viable was the postulated step of ethylene protonation, which was
supposed to result in a highly unstable carbocation 3. As shown in this work, there is another
type of intermediate/transition state for Prins mechanism which does not require
protonation of ethylene. This intermediate was also identified by Yamabe, et al. 77. In their
97
theoretical work, propylene and formaldehyde were reacted via a novel C4 intermediate.
Finally, other evidences of reaction between olefin and aldehyde were shown in US Patent
no 2377025 A for 1,3-butadiene production, albeit through an acetylene intermediate, on
alumina-silver and Cr- and Mo-oxide catalyst78, isobutylene and pentenes with
formaldehyde on KU-2 cation exchanged resin to make dioxanes 79 and 1,3-butanediol
production via reaction of propylene and formaldehyde over ceria catalysts that contain
mostly Lewis acid sites 80.
Work presented here for the enol formation step also shows a much lower free-
energy barrier, compared to ethanol dehydrogenation, with a much more negative ΔGRx of
reaction, 16 and -11.8 kcal/mol, respectively. Chieregato et al. suggested a novel
mechanism with ethanol releasing a proton from its β-carbon and yielding a carbanion with
~33-36 kcal/mol forward barrier and negligible reverse barrier. This carbanion would then
react with either ethanol or acetaldehyde to yield 1-butanol or crotyl alcohol, respectively,
which subsequently dehydrate to produce 1,3-butadiene41. The carbanion has an interesting
configuration in which the ethanol is not deprotonated, rather the hydroxyl group is
interacting with a proton detached from β-carbon. From their overall mechanism, the rate-
limiting step was predicted to be the reaction of acetaldehyde and carbanion with the
electronic energy barrier of 11.4 kcal/mol with respect to the adsorbed reactants. The
reaction to form C4 hydrocarbons is exergonic. Our attempt on cleaving the proton from
β-carbon, however, lead to another pathway, namely to dehydration to form ethylene in
5A, 5B TS and 5C in Figure 3.2 (ΔGA,forward =33.5 kcal/mol, ΔGRx=-4.9 kcal/mol). The
unstable carbanion situation that would lead to C-O bond scission was not encountered in
the case of diol transformation to 1,3-butadiene (Figure 3.2D-K). As a result, Figure 3.3F
98
shows a C4 molecule with two oxygen atoms bound to the surface. The terminal carbon,
however, is in distorted sp2 configuration and thus, represents a carbanion. Similarly, on
the investigated Prins mechanism, a stabilized C4 carbanion, which leads to the desorption
of 1,3-butadiene, is also observed in Figure 3.5G and 3.5E iii 2.
Interesting observation stemming from our work was that the Prins mechanism for
C-C bond formation was thermodynamically more favorable than aldol C-C coupling step,
and with the calculated barrier of 28.8 kcal/mol, i.e. ~10 kcal/mol lower than ethanol
dehydrogenation. The activation energy is, however, still larger than that of aldol
condensation (16.1 kcal/mol). Another fact complicating our conclusions further is that
adsorption of both C2 intermediates on the surface is almost unfavorable
thermodynamically (adsorption free energy (ΔGAds) = -0.2 kcal/mol), and the transition
state is located above the reference state. This step was similar to the
carbanion/acetaldehyde reaction, which is also a double bond opening of two sp2 carbon
atoms in acetaldehyde and ethylene. The suggested carbanion/acetaldehyde reaction
involved butane-1,3-diol as an intermediate which subsequently deprotonates, as opposed
to but-3-en-2-ol (state 5C) suggested by our calculation. 1,3-butanediol, however, still
appears in our mechanism as a product of MPV reduction on the resulting acetaldol from
aldol condensation.
4. Conclusion
A complex reactive mechanism of ethanol to form 1,3-butadiene was explored
using periodic quantum chemical methods. Overall free energy surface was explored for
99
the highly debated reaction mechanisms, including Toussaint’s aldol condensation
mechanism, Fripiat’s Prins mechanism and mechanism based on Ostromislensky’s
hemiacetal rearrangement. Based on the thermodynamic and kinetic data determined
within this study we identified four rate limiting steps in the overall process. In particular,
ethanol dehydration to form ethylene possessed lower energy barrier than dehydrogenation
to yield acetaldehyde suggesting competing reactive pathways. Aldol condensation step to
form acetaldol is preceded with forward free-energy barrier of 16.1 kcal/mol but limited
thermodynamically with endergonic reaction free energy of 12.9 kcal/mol. This calculation
also offers another viable route in the form of Prins condensation, which has a free energy
barrier of 28.8 kcal/mol with exergonic reaction free energy of -22.7 kcal/mol. Finally,
thermodynamic stability of 1-ethoxyethanol prevents further reaction via hemiacetal
rearrangement. The results presented here provide a first glimpse into the 1,3-butadiene
formation mechanism on undoped MgO reactive sites in light of the vast literature
discussing variety of the proposed mechanistic pathways mostly based on conventional
homogenous organic chemistry reactions. While the surface model employed in this work
utilized most reactive MgO site, presence of H2O as a reaction product suggests that other
surface sites, based on reactive hydroxyls, can also affect the overall reactive pathways and
will be the focus of the future studies. However, based on the present calculations alone
several mechanisms appear possible. Reactivity experiments are needed to discriminate
between the different hypothesis, and we hope that our calculations will stimulate such
studies.
100
References
(1) Matar, S.; Hatch, L. F. Chemistry of Petrochemical Processes; Elsevier, 2001.
(2) McCoy, M.; Reisch, M. S.; Tullo, A. H.; Short, P. L.; Tremblay, J. F.; Storck, W.
J. Chem. Eng. News 2007, 85, 29.
(3) Angelici, C.; Weckhuysen, B. M.; Bruijnincx, P. C. a. ChemSusChem 2013, 6 (9),
1595–1614.
(4) Sun, J.; Wang, Y. ACS Catal. 2014, 4 (4), 1078–1090.
(5) Han, Y.; Lu, C.; Xu, D.; Zhang, Y.; Hu, Y.; Huang, H. Appl. Catal. A Gen. 2011,
396 (1–2), 8–13.
(6) Zhang, X.; Wang, R.; Yang, X.; Zhang, F. Microporous Mesoporous Mater. 2008,
116 (1–3), 210–215.
(7) Meng, T.; Mao, D.; Guo, Q.; Lu, G. Catal. Commun. 2012, 21 (0), 52–57.
(8) Song, Z.; Takahashi, A.; Nakamura, I.; Fujitani, T. Appl. Catal. A Gen. 2010, 384
(1–2), 201–205.
(9) Sun, J.; Zhu, K.; Gao, F.; Wang, C.; Liu, J.; Peden, C. H. F.; Wang, Y. J. Am.
Chem. Soc. 2011, 133 (29), 11096–11099.
(10) Liu, C.; Sun, J.; Smith, C.; Wang, Y. Appl. Catal. A Gen. 2013, 467 (0), 91–97.
(11) Jørgensen, B.; Egholm Christiansen, S.; Dahl Thomsen, M. L.; Christensen, C. H.
J. Catal. 2007, 251 (2), 332–337.
(12) Carotenuto, G.; Tesser, R.; Di Serio, M.; Santacesaria, E. Catal. Today 2013, 203
(0), 202–210.
(13) Birky, T. W.; Kozlowski, J. T.; Davis, R. J. J. Catal. 2013, 298 (0), 130–137.
(14) Tsuchida, T.; Sakuma, S.; Takeguchi, T.; Ueda, W. Ind. Eng. Chem. Res. 2006, 45
(25), 8634–8642.
(15) Lebedev, S. V. Zh Obs. Khim 1933, No. 3, 698–717.
(16) Jones, M. D. Chem. Cent. J. 2014, 8 (1), 53.
(17) Chang, F.-W.; Yang, H.-C.; Roselin, L. S.; Kuo, W.-Y. Appl. Catal. A Gen. 2006,
304 (0), 30–39.
(18) Makshina, E. V.; Janssens, W.; Sels, B. F.; Jacobs, P. A. Catal. Today 2012, 198
(1), 338–344.
(19) De Baerdemaeker, T.; Feyen, M.; Müller, U.; Yilmaz, B.; Xiao, F.-S.; Zhang, W.;
Yokoi, T.; Bao, X.; Gies, H.; De Vos, D. E. ACS Catal. 2015, 3393–3397.
(20) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A.
ChemSusChem 2014, 7 (9), 2505–2515.
(21) Ohnishi, R.; Akimoto, T.; Tanabe, K. J. Chem. Soc. Chem. Commun. 1985, No.
22, 1613–1614.
(22) Sekiguchi, Y.; Akiyama, S.; Urakawa, W.; Koyama, T.; Miyaji, A.; Motokura, K.;
Baba, T. Catal. Commun. 2015, 68, 20–24.
(23) Larina, O.; Kyriienko, P.; Soloviev, S. Catal. Letters 2015, 1–7.
(24) Sushkevich, V. L.; Ivanova, I. I.; Ordomsky, V. V; Taarning, E. ChemSusChem
2014, 7 (9), 2527–2536.
(25) Janssens, W.; Makshina, E. V.; Vanelderen, P.; De Clippel, F.; Houthoofd, K.;
Kerkhofs, S.; Martens, J. A.; Jacobs, P. A.; Sels, B. F. ChemSusChem 2014, 8 (6),
994–1008.
(26) Kvisle, S.; Aguero, A.; Sneeden, R. P. A. Appl. Catal. 1988, 43 (1), 117–131.
101
(27) Kitayama, Y.; Satoh, M.; Kodama, T. Catal. Letters 1996, 36 (1–2), 95–97.
(28) Kitayama, Y.; Michishita, A. J. Chem. Soc., Chem. Commun. 1981, No. 9, 401–
402.
(29) Niiyama, H.; Morii, S.; Echigoya, E. Bull. Chem. Soc. Jpn. 1972, 45 (3), 655–659.
(30) Bhattacharyya, S. K.; Sanyal, S. K. J. Catal. 1967, 7 (2), 152–158.
(31) Bhattachariya, S.; Gangooli, N. D.; Avastkhi, B. N. J. Catal. 1965, 4 (6), 736.
(32) Bhattacharyya, S. K.; Avasthi, B. N. Ind. Eng. Chem. Process Des. Dev. 1963, 2
(1), 45–51.
(33) Sushkevich, V. L.; Ivanova, I. I.; Taarning, E. Green Chem. 2015.
(34) Baylon, R. a. L.; Sun, J.; Wang, Y. Catal. Today 2015, 1–7.
(35) Bhattacharyya, S. K.; Ganguly, N. D. J. Appl. Chem. 1962, 12 (3), 97–104.
(36) Jones, M. D.; Keir, C. G.; Iulio, C. Di; Robertson, R. A. M.; Williams, C. V;
Apperley, D. C. Catal. Sci. Technol. 2011, 1 (2), 267–272.
(37) Quattlebaum, W. M.; Toussaint, W. J.; Dunn, J. T. J. Am. Chem. Soc. 1947, 69,
593–599.
(38) Toussaint, W. J.; Dunn, J. T.; Jachson, D. R. Ind. Eng. Chem. 1947, 39 (2), 120–
125.
(39) Ostromislenskiy, I. J. Russ. Phys. Chem. Soc. 1915, No. 47, 1472 –1506.
(40) Gruver, V.; Sun, A.; Fripiat, J. J. Catal. Letters 1995, 34 (3–4), 359–364.
(41) Chieregato, A.; Velasquez Ochoa, J.; Bandinelli, C.; Fornasari, G.; Cavani, F.;
Mella, M. ChemSusChem 2014, 8 (2), 377–388.
(42) Zhang, M.; Gao, M.; Chen, J.; Yu, Y. RSC Adv. 2015, 5 (33), 25959–25966.
(43) Chizallet, C.; Costentin, G.; Che, M.; Delbecq, F.; Sautet, P.; Surface, D.; Marie,
P. J. Am. Chem. Soc. 2007, No. 19, 6442–6452.
(44) Shalabi, A. S.; El-Mahdy, A. M. Phys. Lett. A 2001, 281 (2–3), 176–186.
(45) Chizallet, C.; Bailly, M. L.; Costentin, G.; Lauron-Pernot, H.; Krafft, J. M.; Bazin,
P.; Saussey, J.; Che, M. Catal. Today 2006, 116 (2), 196–205.
(46) Di Valentin, C.; Del Vitto, A.; Pacchioni, G.; Abbet, S.; Wörz, A. S.; Judai, K.;
Heiz, U. J. Phys. Chem. B 2002, 106 (46), 11961–11969.
(47) Branda, M. M.; Rodríguez, A. H.; Belelli, P. G.; Castellani, N. J. Surf. Sci. 2009,
603 (8), 1093–1098.
(48) Kantorovich, L. N.; Holender, J. M.; Gillan, M. J. Surf. Sci. 1995, 343 (3), 221–
239.
(49) Sushko, P. V.; Shluger, A. L.; Catlow, C. R. A. Surf. Sci. 2000, 450 (3), 153–170.
(50) Chizallet, C.; Petitjean, H.; Costentin, G.; Lauron-Pernot, H.; Maquet, J.;
Bonhomme, C.; Che, M. J. Catal. 2009, 268 (1), 175–179.
(51) Aramendia, M. a.; Borau, V.; Jimenez, C.; Marinas, J. M.; Porras, A.; Urbano, F. J.
J. Mater. Chem. 1996, 6 (12), 1943.
(52) Bailly, M.; Chizallet, C.; Costentin, G.; Krafft, J.; Lauronpernot, H.; Che, M. J.
Catal. 2005, 235 (2), 413–422.
(53) Cornu, D.; Guesmi, H.; Krafft, J. M.; Lauron-Pernot, H. J. Phys. Chem. C 2012,
116 (11), 6645–6654.
(54) Corson, B. B.; Jones, H. E.; Welling, C. E.; Hinckley, J. A.; Stahly, E. E. Ind. Eng.
Chem. 1950, 42 (2), 359–373.
(55) Zhang, M.; Yu, Y. Ind. Eng. Chem. Res. 2013, 52 (28), 9505–9514.
(56) Redina, E. A.; Greish, A. A.; Mishin, I. V; Kapustin, G. I.; Tkachenko, O. P.;
102
Kirichenko, O. A.; Kustov, L. M. Catal. Today 2015, 241, Part, 246–254.
(57) Fessenden, R. J.; Fessenden, J. S. In Organic Chemistry; Dodson, K., Ed.;
Brooks/Cole Publishing Company, 1995; pp 713–748.
(58) Harmony, M. D.; Laurie, V. W.; Kuczkowski, R. L.; Schwendeman, R. H.;
Ramsay, D. A.; Lovas, F. J.; Lafferty, W. J.; Maki, A. G. J. Phys. Chem. Ref. Data
1979, 8 (3), 619.
(59) Yang, Y.-J.; Teng, B.-T.; Liu, Y.; Wen, X.-D. Appl. Surf. Sci. 2015, 357, Part,
369–375.
(60) Chen, H.-Y. T.; Giordano, L.; Pacchioni, G. J. Phys. Chem. C 2013, 117 (20),
10623–10629.
(61) Cavalleri, M.; Pelmenschikov, A.; Morosi, G.; Gamba, A.; Coluccia, S.; Martra, G.
In Oxide-based Systems at the Crossroads of ChemistrySecond International
Workshop October 8-11, 2000, Como, Italy; A. Gamba, C. C. and S. C. B. T.-S. in
S. S. and C., Ed.; Elsevier, 2001; Vol. Volume 140, pp 131–139.
(62) Barborini, M.; Guidoni, L. J. Chem. Phys. 2012, 137 (22).
(63) Bučko, T.; Benco, L.; Hafner, J.; Ángyán, J. G. J. Catal. 2011, 279 (1), 220–228.
(64) Fessenden, R. J.; Fessenden, J. S. In Organic Chemistry; Dodson, K., Ed.;
Brooks/Cole Publishing Company, 1995; pp 275–308.
(65) Musolino, V.; Selloni, A.; Car, R. J. Chem. Phys. 1998, 108 (12), 5044.
(66) Pacchioni, G. J. Chem. Phys. 1996, 104 (18), 7329.
(67) Zhukovskii, Y. F.; Kotomin, E. a.; Borstel, G. Vacuum 2004, 74 (2), 235–240.
(68) Matveev, A.; Neyman, K.; Yudanov, I.; Rösch, N. Surf. Sci. 1999, 426 (1), 123–
139.
(69) Rodriguez, J. a.; Jirsak, T.; Chaturvedi, S. J. Chem. Phys. 1999, 111 (17), 8077.
(70) Rodriguez, J. a.; Jirsak, T.; Freitag, A.; Larese, J. Z. J. Phys. Chem. B 2000, 2
(100), 7439–7448.
(71) Angelici, C.; Meirer, F.; van der Eerden, A. M. J.; Schaink, H. L.; Goryachev, A.;
Hofmann, J. P.; Hensen, E. J. M.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS
Catal. 2015, 5 (10), 6005–6015.
(72) Boronat, M.; Corma, A.; Renz, M. J. Phys. Chem. B 2006, 110 (42), 21168–21174.
(73) Pasini, T.; Lolli, A.; Albonetti, S.; Cavani, F.; Mella, M. J. Catal. 2014, 317, 206–
219.
(74) Assary, R. S.; Curtiss, L. A.; Dumesic, J. A. ACS Catal. 2013, 3 (12), 2694–2704.
(75) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A. Catal.
Sci. Technol. 2015.
(76) Arundale, E.; Mikeska, L. A. Chem. Rev. 1952, 51 (3), 505–555.
(77) Yamabe, S.; Fukuda, T.; Yamazaki, S. Beilstein J. Org. Chem. 2013, 9, 476–485.
(78) Miller, H. Conversion of a mixture of acetaldehyde and ethylene to butadiene.
US2377025 A, 1945.
(79) Isagulyants, V. I.; Safarov, M. G. Pet. Chem. U.S.S.R. 1966, 5 (3), 203–208.
(80) Wang, Y.; Wang, F.; Song, Q.; Xin, Q.; Xu, S.; Xu, J. J. Am. Chem. Soc. 2013,
135 (4), 1506–1515.
(81) Towns, J.; Cockerill, T.; Dahan, M.; Foster, I.; Gaither, K.; Grimshaw, A.;
Hazlewood, V.; Lathrop, S.; Lifka, D.; Peterson, G. D.; Roskies, R.; Scott, J. R.;
Wilkins-Diehr, N. Comput. Sci. Eng. 2014, 16 (5).
103
Chapter 4
Surface Reaction Mechanisms Study of
MgO/SiO2 for Lebedev Process
Abstract .......................................................................................................................... 103
1. Introduction ....................................................................................................... 104
2. Results and Discussion ...................................................................................... 109
2.1. Catalyst activity and selectivity testing ....................................................... 109
2.2. In-situ DRIFT spectroscopy of MgO based catalyst surface hydroxyl
groups................................................................................................................... 110
2.3. Acid-base characterization of WK (1:1) catalyst using CO2 and pyridine
as probe molecules .............................................................................................. 113
2.4. In-situ DRIFT spectroscopy to monitor hydroxyl group reactivity during
the ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol adsorption and
subsequent reaction on a WK (1:1) catalyst surface ........................................... 115
2.5. In-situ DRIFT spectroscopy of C2 (ethanol, acetaldehyde) and C4
(crotonaldehyde and crotyl alcohol) adsorption and reaction on WK (1:1)
catalyst surface as a function of temperature ...................................................... 119
2.5.1. C2 reactants and intermediates ............................................................... 119
2.5.2. C4 intermediates ..................................................................................... 127
2.6. DFT calculations ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol
vibrational frequencies ........................................................................................ 130
2.7. In-situ DRIFT spectra for the ethanol, acetaldehyde, crotonaldehyde and
crotyl alcohol reaction on a WK (1:1) catalyst surface: the effect of the vapor
phase presence .................................................................................................... 136
3. Conclusions ........................................................................................................ 145
Supporting Information ............................................................................................... 148
References ...................................................................................................................... 151
Abstract
1,3-butadiene is an important commodity chemical and new, selective routes of
catalytic synthesis using green feedstock, such as ethanol, is of interest. For this purpose,
surface chemistry of MgO/SiO2 catalyst synthesized using wet-kneading was explored
during the reaction of ethanol and the corresponding reactive intermediates, including
104
acetaldehyde, crotonaldehyde, crotyl alcohol using temperature programmed in situ
DRIFT spectroscopy combined with DFT calculations. Ethanol adsorption yielded several
physisorbed and chemisorbed surface species. Acetaldehyde exhibited high reactivity to
form crotonaldehyde. However, aldehyde intermediates resulted in strongly bound surface
species stable even at high temperatures, assigned to surface acetate, and/or 2,4-hexadienal
or polymerized acetaldehyde. Crotonaldehyde was reduced by ethanol to yield crotyl
alcohol via MPV mechanism. Crotyl alcohol, on the other hand, showed to be very reactive
and yield two different species on the surface, namely physisorbed and deprotonated that
would further desorb as 1,3-BD. Presence of gas phase hydrogen containing molecules,
such as ethanol, proved to be key in several reactive steps, including acetaldehyde
condensation step and crotonaldehyde reduction. Altogether, these data suggested a
complex reactive interactions between the surface hydroxyl groups, gaseous reactants and
surface bound reactive intermediates during the 1,3-BD formation. Future work is needed
to correlate vapor phase product evolution with the transient reactive surface intermediates
to examine trends leading to higher overall 1,3-BD selectivity.
1. Introduction
1,3-butadiene (1,3-BD) is an important commodity chemical with widespread
applications in polymer synthesis and as an organic chemistry intermediate.1 It is
commonly produced via crude oil cracking. 1,3-BD production can be affected by market
instability triggered by oil price fluctuations. This difficulty is compounded by the
emergence of shale gas, which suggests a need for an alternative and more sustainable
production method.2 For this reason, there has been a renewed interest in utilizing ethanol
105
as a feedstock for 1,3-BD synthesis. Production from ethanol was used during World War
II by the USA and the USSR with a two-step process and a one-step catalytic process,
respectively, as demonstrated by Ostromislensky and Lebedev.3,4 Several reports since then
have highlighted the economic viability of the overall synthesis process with the one-step
catalytic process recently becoming a focal point.1,2,5 This single step process originally
utilized MgO/SiO2 catalyst with a 30-40% yield. Ethanol dehydrogenation to yield
acetaldehyde was identified as the rate-determining step in the generally accepted complex
reaction mechanism,6 with a large body of experimental work performed for elucidating
the reaction pathways.7–12 However, there is only a limited number of studies that focus
on the adsorbed reactive surface intermediates on MgO catalysts that utilize in-situ
spectroscopy to characterize the reaction intermediates under operating conditions.8,10,13,14
A one-step catalytic mechanism, as summarized in Figure 4.1, involves
dehydrogenation of ethanol to acetaldehyde on basic MgO sites1,10,15, followed by C-C
I II II
IV V VI
Figure 4.1. Main reaction mechanism proposed for ethanol to 1,3-butadiene via
Toussaint’s aldol condensation.
106
coupling via the aldol condensation mechanism to yield crotonaldehyde.6,16,17
Crotonaldehyde can be further hydrogenated via MPV (Meerwein–Ponndorf–Verley)
reduction, either with ethanol or molecular H2 and the resulting crotyl alcohol dehydrated
to give 1,3-BD.16,18 Three computational studies by Chieregato et al. 10, Zhang et al.11 and
Taifan et al.17 attempted to unravel the reaction mechanism and the structures of the
reactive surface intermediates. Zhang et al. performed calculations using periodic density,
functional generalized, gradient approximation (GGA), with a focus on the very first step
of the overall reaction mechanism, the dehydrogenation of ethanol. A stepped MgO
surface was predicted to have a lower energy barrier than a flat surface for this reaction.11
The dissociation of ethanol on that surface was studied for three different surface sites, i.e.,
Mg5CO5C, Mg4CO4C, and Mg3CO3C and on the stepped surface, i.e. Mg4CO4C was shown to
have the lowest potential energy barrier for this reaction. Chieregato et al., on the other
hand, proposed an entirely different mechanism based on cluster type calculations and a
Gaussian basis set. They ruled out crotonaldehyde and crotyl alcohol as possible
intermediates and concluded that acetaldehyde would react with a carbanion, which
resulted from ethanol C-H bond cleavage.10 Taifan et al., for the first time, outlined a
complete reactive pathway for ethanol conversion to 1,3-BD by using periodic GGA
calculations. The pathways explored included an alternative Fripiat’s Prins mechanism and
a mechanism based on Ostromislensky’s hemiacetal rearrangement. They showed ethanol
dehydration to have an energetic barrier comparable to that of dehydrogenation. The
dehydration proceeded on Mg3c with surface O5c responsible for the initial proton transfer
and the resulting low coordination O2c and O4c hydroxyl group formation.
Dehydrogenation, on the other hand, took place in the vicinity of Mg3c and step O4c.
107
Acetaldol (3-hydroxybutanal) formation proceeded via a 16 kcal/mol free energy barrier,
as calculated at 450 oC using harmonic/rigid rotor approximation. That took place when
acetaldehyde molecules were coordinated to Mg3c, Mg4c and Mg5c surface sites. Acetaldol
was not identified as a stable intermediate on a PES surface, as it immediately deprotonated
on reactive O4c sites. While computational modelling utilized low saturation Mg3c and O4c
sites for the overall reaction cycle, spectroscopic identification of the corresponding sites
and the reactive surface intermediates was less utilized. In particular, an adsorbed ethoxy
group on MgO was identified at 1119-1132 cm-1 in the temperature regime between 200
and 400 oC, when adsorbed ethanol was heated suggesting ethanol chemisorption.10 New
bands appeared at 1718 and 1143 cm-1 already at 150 oC and these bands were attributed
to acetaldehyde and new C-O containing surface species assigned to carbanion,
respectively. A transient peak at 1653 cm-1 together with one at 2957 cm-1 were assigned
to acyl or acetyl species. The reactive adsorbed intermediate at 1620 cm-1 was observed
and assigned to crotyl alcohol.19 C=O and C=C stretching vibrations at 1672 and 1649 cm-
1 observed above 350 oC were assigned to other C4 products, such as 1,3-BD,
crotonaldehyde and butanol. No adsorbed crotonaldehyde or acetaldol intermediates (the
latter one in agreement with Taifan et al.)17 were observed at lower (>250 oC) temperatures
but 1,3-BD formation was identified, suggesting that the aldol condensation mechanism
was not a key. In-situ DRIFTS was also the preferred technique used by Davis’ group13,14
and Ordomsky et al.8 to monitor the surface species during the reaction to n-butanol and
1,3-butadiene, respectively. Ethanol strongly adsorbed on MgO as both dissociated
ethoxide and a molecularly adsorbed ethanol. Dissociated ethanol exhibited a major peak
at 1132-1119 cm-1, while molecularly adsorbed ethanol was detected at 1058 cm-1. At
108
higher temperatures, no aldol condensation was detected during the experiment, possibly
due to the very low conversion which was also supported by a small acetaldehyde formed
at 1711 cm-1.13,14 Aldol condensation of acetaldehyde was studied over MgO/SiO2 catalyst
and was suggested to instantaneously take place on the surface once acetaldehyde was
introduced to the IR cell, as shown by the band at 1643 cm-1, attributed to C=C stretch of
crotonaldehyde. Other than aldol condensation, acetaldehyde undergoes other side
reactions, namely condensation on the basic sites, as well as aldol condensation with
crotonaldehyde yielding 2,4-hexadienal, an unsaturated aldehyde.8
In this work we report a detailed study of wet-kneaded MgO/SiO2 catalyst surface
reactive sites and reactive intermediates during the ethanol conversion to 1,3-BD. Where
necessary, data are also reported for pure MgO model catalyst. Wet kneading (WK) of
MgO/SiO2 catalyst has been shown to result in an active catalysis towards 1,3-BD
formation from ethanol.1,15,20–23 In this study, we prepared calcined MgO and MgO/SiO2
catalysts using a wet kneading method with an MgO:SiO2 mass ratio of 1:1. We used in
situ Diffuse Reflectance Infrared Spectroscopy (DRIFTS) and the corresponding proposed
reactants and proposed reactive intermediates, including ethanol, acetaldehyde, crotyl
alcohol, crotonaldehyde, as shown in Figure 4.1. To aid in the observed peak assignment,
quantum chemical calculations using periodic boundary conditions and PBE functional
were used.
109
2. Results and Discussion
2.1. Catalyst activity and selectivity testing
Figure 4.2 depicts the catalytic activity testing of the synthesized wet-kneaded
MgO/SiO2 catalyst, WK (1:1). The experiment were carried out at 723 K (450°C) at several
WHSV (hr-1) ranging from 0.78 to ~2 hr-1. WHSV plays a very important role in
determining the catalyst’s activity, since it represents the catalyst-to-reactant ratio. The
conversion decreased with increasing WHSV from ~87% to ~60%. Selectivities of selected
products, i.e. acetaldehyde, ethylene, and 1,3-butadiene, were relatively unaffected by
WHSV. At very high conversion, there were several other byproducts, such as butenes,
propene, ethers and some aromatic compound that coked the catalyst; this led to carbon
balance of 60-80%. High 1,3-butadiene selectivity was achieved with this catalyst, 35-
40%, without the addition of routinely used transition metal oxide promoter.1,2 Similar
conversion-yield values were previously reported by Weckhuysen’s group, where different
methods of preparation and precursors were explored to find the best working catalyst.21
Table 4.1 shows comparison between catalyst in this work and previously used wet-
kneaded MgO/SiO2 catalysts.
Table 4.1. Catalytic activity comparison of WK (1:1) with previously investigated wet-
kneaded synthesized catalysts.
Catalyst T (K) WHSV (h-1) XEtOH
(%)
YBD
(%)
PBD
(gBD g-1cat h-1) Ref
WK (1:1) 723 1.1 ~84 33 0.4 This
work
WK-a 623 0.15 50 42 0.06 Makshina
et al.23
WK-b 698 1.1 ~67 35 0.25 Angelici
et al.21
110
2.2. In-situ DRIFT spectroscopy of MgO based catalyst surface hydroxyl groups.
We begin by investigating the hydroxyl groups present on a dehydrated WK (1:1)
surface by comparing them with MgO and SiO2. In-situ IR spectroscopy results for the
pure MgO surface shown in Figure 4.3 were obtained after heating (dehydrating) the
sample at 773 K, typical for the ethanol catalytic reaction to form 1,3-BD, in air and cooling
down to 373 K temperature. Spectra show two high basicity (low coordination) peaks in
the hydroxyl region at 3765 and 3745 cm-1, while several broad peaks are also present at
3700-3400 cm-1, namely, 3660, 3547 and 3465 cm-1. The higher stretching frequency is
related to a more isolated (and basic) hydroxyl group, while the lower one is often assigned
to multi-coordinated hydrogen bonded hydroxyls.27,39–43 In general, there have been six
structural hydroxyl group models proposed to exist on MgO. Anderson et al. proposed two
kinds of hydroxyl groups on the MgO surface: hydrogen bond acceptor and hydrogen bond
donor.41 Their model was subsequently refined by Shido et al., where the two regions
Figure 4.2. Conversion (●) and selectivity of main products (■ acetaldehyde; ▲
ethylene; ♦ 1,3-butadiene) at different WHSV. Reaction conditions: T=723 K, Qtot = 50
cm3/min, Mcat=0.2 g, P0EtOH = 2.72; 3.77; 5.15; 6.96 kPa.
111
could be classified in further detail based on the coordination numbers of the Mg and O
atoms.42 Coluccia, Morrow, and Knozinger each proposed three different models with one
uniting characteristic: the inclusion of an isolated hydroxyl group as the sharp band at the
high wavenumber region and hydrogen bond donor - multicoordinated hydroxyl groups in
the lower wavenumber region.39,40,43 Most recent models combined DFT and infrared
spectroscopy studies to show that the most isolated single coordinated (O1C-H) group does
not exist: it immediately transforms into O3C-H and O4C-H at a lower temperature and into
O2C-H at an elevated desorption temperature, and thus a new model was proposed.44 The
bands observed at 3765 and 3745 cm-1 in our work agree well with those reported in the
literature. Those bands have been assigned to low coordinated O1c-H or O2c-H hydrogen
bond acceptors or O4c-H and O5c-H coordinated isolated groups on valleys and edges of
the MgO crystallites.39,44 The peaks below 3650 cm-1 are in general attributed to multi-
coordinated hydrogen bond donor hydroxyl groups44, thus presenting a rather complex
picture of the reactive MgO surface.
Figure 4.3. In-situ DRIFTS spectra acquired of dehydrated (temperature programmed
to 773 K at 10 oC/min under air and cooled down to 100 oC) MgO, MgO WK (1:1)
catalysts and SiO2. Only hydroxyl region of 3800 to 3200 cm-1 is shown. Spectra are
acquired at 100 °C.
112
The WK (1:1), on the other hand, exhibited four major peaks at 3745, 3725, 3705,
and 3680 cm-1. For comparison, dehydrated spectra of calcined SiO2 are also shown. There
are two bands present for SiO2, a sharp one at 3745 cm-1 and a broad band at 3700-3450
cm-1. The sharp peak is typical for the isolated silanol (Si-OH) vibration with a small
contribution from the geminal silanol group (HO-Si-OH), while the broad band is formed
from the contribution of the hydrogen bonded vicinal silanol groups.45 The 3745 cm-1 peak
is also observed in WK (1:1), albeit at the lower intensity, which suggests that isolated
silanol groups are consumed during the wet-kneading interaction with MgO. The other
three peaks present, 3725, 3705, and 3680 cm-1, are unique to the WK (1:1) structure. The
latter peak has previously been assigned to magnesium silicates, due to its formation in the
presence of silica.7,19,46 It has previously been observed as a mineral lizardite hydroxyl
group at 3686 cm-1.47 The relatively low FWHM (Full Width at Half Maximum) of the
peak suggests that this group might be isolated, rather than hydrogen bonded, consistent
with the crystalline structure of the lizardite.48 Peaks at 3725 and 3705 cm-1 are more
difficult to assign directly, since none of the magnesium silicate compounds exhibit
hydroxyl stretches above 3700 cm-1.47 It can be proposed that the interaction of MgO and
SiO2 during wet kneading increases the formation of hydroxyl groups that are already
present on MgO itself, i.e., wet-kneading results in more defects that produce the said
hydroxyl group or that peaks could originate from Mg-OH interacting with nearby SiO2
surface sites. The peak at 3725 cm-1 is rather intriguing due to the fact that it was not
observed by other groups. We tentatively assign the peaks at 3725 and 3705 cm-1 to the
isolated O4c-H and O5c-H coordinated groups formed in the presence of the amorphous
SiO2 (SiMg4cO4c and SiMg4cO5c). This is also consistent with the decrease in intensity of
113
the 3765 and 3745 cm-1 hydroxyl groups present on MgO but not on WK (1:1), where
incorporation of amorphous Si-O-Mg linkages could result in the frequency shift towards
lower wavenumbers.
2.3. Acid-base characterization of WK (1:1) catalyst using CO2 and pyridine as
probe molecules
Characterization of the basic sites present on WK (1:1) was carried out by
performing in-situ DRIFTS using CO2 as a probe molecule. Figure 4.4a depicts the spectra
of adsorbed CO2 species at different temperatures. The basicity was previously reported to
originate from MgO, and with no contribution from SiO2.2 There are three broad, main
peaks present on the spectra, located at 1650, 1531, and 1405 cm-1. Judging from the
carbonate υ3 frequency split, the last two peaks originate from monodentate carbonate,
assigned to υ3 as and υ3 s, respectively.49 The peak at 1650 cm-1 could originate from either
bidentate carbonate or monodentate bicarbonate. However, bicarbonate would exhibit a
peak at around ~1250-1200 cm-1, which is non-existent in this case. Bidentate carbonate
assignment is more likely than bicarbonate, given the broad peaks exhibited in this spectra,
where the accompanying υ3 s would be convoluted as a shoulder to the peak at 1405 cm-1.
Furthermore, the basic site strength can be determined by the surface species present.
Monodentate carbonate is typically more stable than bidentate carbonate, while bicarbonate
is the least stable.2,49 Hence, the strong basic sites are assigned to monodentate carbonate,
while medium-strength and weak basic sites are assigned to bidentate carbonate and
bicarbonate, respectively. From the spectra, the bidentate carbonate is far less intense than
monodentate carbonate, indicating the more basic nature of the MgO/SiO2 WK (1:1).
Several other methods of preparation, such as sol-gel19 and incipient wetness
114
impregnation2, yield catalysts with limited amount of strong basic sites. However, it should
be noted that at the reaction temperature, ~673-723 K, the CO2 species in our catalyst are
mostly absent. This indicates that the basic sites present on the catalyst gradually lose
strength at elevated temperature.
The acidity of the catalyst was characterized using pyridine as the probe molecule
(Figure 4.4b). Peaks at 1445 and 1605 cm-1 indicates the presence of strong Lewis acid
sites, while the peak at 1577 cm-1 is for weak Lewis acid sites.50 Peak at 1490 cm-1 is
assigned to a combination band of Lewis and Brønsted acid sites, while Brønsted acid site
itself should exhibit a peak at 1540 cm-1, which is not present on our catalyst. The peak at
1595 cm-1 does not represent any acid sites, instead, it was assigned to hydrogen-bound
pyridine.50 The absence of Brønsted acid sites were also observed by previous
investigators, given the basic nature of the catalyst.2,7,19,51 However, the intensity of the
strong Lewis acid sites is a dominant feature on this spectra indicating that the catalyst
possess a significant amount of strong Lewis acid sites, relative to the weaker Lewis acid
sites. SiO2 by itself is known to be slightly acidic, contributing to the weak Lewis acid
(a) (b)
Figure 4.4. DRIFTS spectra of adsorbed (a) CO2 and (b) pyridine on WK (1:1) catalyst
at different temperatures to probe the catalyst’s basicity and acidity at relevant
temperatures.
115
sites, while the rest of the Lewis acid sites are combination of defect sites of MgO and the
interaction between SiO2 and MgO.1,2,10,51
2.4. In-situ DRIFT spectroscopy to monitor hydroxyl group reactivity during the
ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol adsorption and subsequent
reaction on a WK (1:1) catalyst surface.
Spectra for those hydroxyl groups in the 3800 to 3200 cm-1 regions during WK
(1:1) reaction with ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol are shown in
Figure 4.5 and tabulated in Table 2. Subtracted adsorbed molecule spectra as a function
of temperature are shown in black, while red dotted spectra are for 373 K hydroxyl groups
reacting upon vapor phase molecule adsorption with catalyst spectrum subtracted.
Notably, the catalyst sample surface was treated at 773 K beforehand; thus, the hydroxyl
groups observed in Figure 4.5 are transient reactive groups formed and released during the
organic molecule adsorption/reaction. Upon adsorption of organic molecules, negative
peaks appeared on all the assigned WK (1:1) hydroxyl groups, i.e. 3747, 3725, 3705, and
3680 cm-1. The adsorption behavior is very different for alcohols – ethanol and crotyl
alcohol – and for aldehydes – acetaldehyde and crotonaldehyde, as shown by the different
intensities of the negative peaks. The alcohols have less affinity to the peaks at 3705 and
3680 cm-1, while aldehydes have no preference on which hydroxyl group to coordinate.
Alcohols’ interactions with MgO surfaces include both molecular adsorption on native
hydroxyl groups52, as well as their displacement via chemisorption, which involves basic
site – Lewis acid site pairs, that will produce adsorbed water as the byproduct.19,53 Positive
hydroxyl peaks in the alcohol cases can indicate new hydroxyl vibrations, due to the newly
116
formed groups via displacement, while the increased hydrogen bonding demonstrates that
some of the native hydroxyl groups only weakly-bind the molecular ethanol. Aldehyde
adsorption, on the other hand, typically takes place via two surface species: on a surface
hydroxyl group via an unstable, protonated intermediate and on a lone pair of oxygen atoms
as a more stable species, typically indicated by the red-shifted C=O stretching vibration at
1650-1680 cm-1.54 Figure 4.5 shows that acetaldehyde adsorbs differently from
crotonaldehyde, that the peak at 3725 cm-1 is not significantly consumed, as compared to
that of crotonaldehyde. We assume that this is due to crotonaldehyde’s π-electron cloud,
which makes the molecule more activated toward consuming the hydroxyl group related
to the 3725 cm-1 peak. For all experiments, the adsorption results in the positive peak at
~3684 cm-1, indicating different hydroxyl group coordination, or a more intense hydrogen
bonding. This positive peak is more intense for acetaldehyde and crotonaldehyde, possibly
due to the formation of new alcoholic species, rather than in the case of alcohols, which
are simply hydroxyl groups displacement.
Table 4.2. Surface hydroxyl group vibrational frequencies during ethanol, acetaldehyde,
crotonaldehyde and crotyl alcohol adsorption on WK (1:1).
Experimental (cm-1)
Ethanol Acetaldehyde Crotonaldehyde Crotyl
alcohol
Assignment
ν
(Mg-
OH)
3748
(3761)
3725
(3721)
-
3680
3755 (3757)
3721
3710
3680
3740
3721
3710
3680
3751
3723
-
3678
Mg4cO4c
SiMg4cO4c
SiMg4cO5c
Mg3Si3O5(OH)4
According to Figure 4.3, the peak at 3747 cm-1 is a combination of both a silica
isolated silanol peak and the basic MgO hydroxyl group. As the temperature is increased,
the former sharply loses intensity, while the latter slowly gains intensity. This trend is true
117
for all the intermediates adsorbed on the surface. Furthermore, this MgO peak splits at a
higher temperature, indicating the presence of a second peak, at lower wavenumber, which
translates to higher coordination. This further splitting was previously observed by
Knözinger et al.39 All other surface hydroxyl groups undergo a significant decrease in
intensity, while also being accompanied by the emergence of their shoulders at a lower
wavenumber as the temperature is increased. One intriguing observation is that those
neighboring hydroxyl peaks are all red-shifted from the native hydroxyl peaks. The thermal
effect on the surface seems to rearrange the hydroxyl group coordination to achieve more
thermodynamically stable configurations, i.e., there are no new hydroxyl groups being
formed.
Putting the rearrangement of the hydroxyl groups aside, increasing the temperature
also led to desorption of the surface species. The release of the hydroxyl groups can be
explained by the flattening shoulder at ~3684 cm-1. These hydroxyl groups were made
during the alcohol/acetaldehyde adsorption. However, native hydroxyl peaks that were
consumed during the initial adsorption keep decreasing in intensity as well. This
continuous decrease indicates that these peaks, in particular at 3747 and 3680 cm-1, are not
fully consumed during the adsorption, i.e. they are relatively less reactive. The remaining,
unconsumed hydroxyl groups of these types undergo further thermal change by achieving
thermodynamically more stable coordination, shown by the increase of the neighboring
hydroxyl peak. For the case of aldehydes, increasing the temperature would both convert
the unstable protonated intermediate into the more stable compound, which is coordinated
to Lewis acid sites.
118
Figure 4.5. In-situ DRIFTS spectra in the hydroxyl group region of 3800 – 3200 cm-1
acquired of ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol on WK (1:1)
catalyst. Sample vapor was adsorbed on the sample surface and temperature ramped
up from 373 to 723 K while spectra being recorded. In-situ DRIFTS dehydrated
catalyst spectrum at 100 °C was used as a reference.
Ethanol Acetaldehyde
Crotonaldehyde Crotyl alcohol
119
2.5. In-situ DRIFT spectroscopy of C2 (ethanol, acetaldehyde) and C4
(crotonaldehyde and crotyl alcohol) adsorption and reaction on WK (1:1) catalyst
surface as a function of temperature.
2.5.1. C2 reactants and intermediates.
When ethanol was adsorbed on WK (1:1), several peaks in the C-H region were
observed between 2985 and 2907 cm-1, as well as at 2720 cm-1, as shown in Figure 4.6 and
tabulated in Table 4.3. The former peaks were attributed to the combination of (CH3)
and (CH2) vibrational modes with the corresponding bending modes located at 1454 and
1418 cm-1, while the assignment of the latter peak is not straightforward. While generally
within the (CH3) spectral region, it can’t be unambiguously assigned. As later shown by
DFT calculations, that peak can be assigned to a frustrated (OH) mode of dissociated
(chemisorbed) ethanol on a Mg4c site. Peaks at 1380 and 1338 cm-1 can be assigned to the
wagging modes of CH2 and CH3. Interestingly, a peak at 1624 cm-1 was observed at 373
K together with a negative peak at ~1670 cm-1 after physisorbed ethanol adsorption. The
peak at 1670 cm-1 is the native bidentate carbonate asymmetric ν3 vibration that persisted
during the sample treatment, which we propose to be displaced due to the reactive ethanol
adsorption on the surface.49 This bidentate carbonate peak would typically be accompanied
by its asymmetric counterpart at ~1370 cm-1,49 however, this spectral region also shows
adsorbed ethanol vibrations. The splitting of the ν3 vibration is typically used to identify
the coordination of the carbonate, since the degree of symmetry lowering caused by surface
coordination is well-known to split the vibration differently.55 For instance, the
monodentate split is ~100 cm-1 (1415 split to ~1400 and ~1500 cm-1), bidentate split is
~300 cm-1, and bridged carbonate split is ≥ 400 cm-1.55 While surface carbonates are
120
typically stable under low O2 and high CO2 concentration, they are typically unstable under
reaction conditions.56 DFT study also showed that ethanol adsorption is
thermodynamically more stable than that for CO2, with ED = 10.5 - 13.5 kcal/mol and 7.1
- 9.4 kcal/mol, respectively,17,57 hence suggesting that chemisorption of ethanol can
displace surface carbonate species. There are several possibilities in assigning the peak at
1624 cm-1. At this wavenumber region, hydrocarbons exhibit C=C stretches, as well as a
distorted C=O stretch. Stable bands at ~1580 cm-1 are also expected for the surface acetate,
even though an accompanying peak must be present around ~1400 cm-1 It is unlikely, that
acetate is formed at 373 K.58 One possible explanation for this is the formation of the
coadsorbed water, presumably formed by dissociation of ethanol that results in the
rearrangement of the hydroxy group when ethanol is adsorbed on the surface. The
formation of water on the surface was previously observed by Busca’s group on γ-Al2O3,
where vapor-phase water was formed right after ethanol was introduced to the surface.53
The corresponding ν(C-O) vibration at ~1100 cm-1 was not observed because of
the predominant vibrations of SiO2 in the spectral region of 1200 to 900 cm-1. However,
peaks at 1140, 1126, 1104 and 1058 cm-1 were observed at 373 K on pure MgO samples,
as shown in the Figure 4.6 inset. The peaks at 1140, 1104, and 1058 can be assigned to
the ethanol species on the surface, both chemisorbed surface ethoxide and physisorbed
ethanol, with spectral shifts observed due to the difference in interaction with and
adsorption on various accessible sites. The shifts between the physisorbed, chemisorbed,
and vapor-phase ethanol were also previously observed by Branda and Birky.14,30 Peaks at
1104 and 1058 cm-1 were assigned to δ(OH) and ν (C-O) of surface ethoxy, respectively,
as also observed by Davis and Cavani groups.10,13,14 The peak at 1140 cm-1 gained intensity
121
after the cell was evacuated and heated to 476 K. This peak was previously assigned to ρw
(CCO) by Davis’ group with the peak located at 1122 cm-1 on MgO.14 Notably, the peak at
1126 cm-1 appears at a higher temperature, ~473 K, which indicates the formation of a new
species. This peak can be assigned to C-C stretch of the adsorbed acetaldehyde.59
Table 4.3. Vibrational frequencies and their assignments for ethanol, acetaldehyde,
crotonaldehyde and crotyl alcohol adsorption on WK (1:1)
Assignment
Experimental (cm-1)
Ethanol Acetaldehyde Enolate Crotonaldehyde Crotyl
alcohol
ν (OH) 2720 - - - -
ν (CH3) 2985 3037, 2967,
2935 - 2967, 2935
3017,
2965
ν (CH2) 2937,
2907, 2881 - - -
2949,
2840
Fermi CH3 2882, 2845
ν (CH) - 2743 - 3032, 2882,
2845
ν (C=O) - 1716, 1680,
1633 -
1716, 1680,
1670 -
ν (C=C) - - 1600,
1578 1600, 1574 1602
δ (CH2) 1454 - - - 1380
δ (CH3) 1418 - - 1456, 1434 1368
ρw (CH) 1380 - - - -
ρw (CH2) - - - - 1441
ρw (CH3) 1338 1456, 1434,
1382 - 1346 1456
ρw (CHO) - 1284 - 1382 -
As the temperature is increased to 723 K, the aforementioned bands start decreasing
in intensity, giving rise to several new bands, including those at 2958, 1653, 1604 and 1581
cm-1. The experiment was carried out under inert gas flow, which prevents chemistry
beyond dehydrogenation and dehydration from happening. The inert atmosphere
encouraged desorption to take place rather than surface reaction, further limiting the
122
reaction to dehydration and dehydrogenation, which are elementary reactions of ethanol.
Experiments under constant reactant flow, where ethanol is constantly supplied to the
surface, were also done and discussed thoroughly in Section 3.7. The first two peaks
appeared intermittently from 473 K up to 623 K. The peak at 1653 cm-1 can be assigned to
the ν (C=O) of acetaldehyde59 adsorbed on an Mg3c surface site with the assignment
confirmed by our DFT calculation. This peak was also previously assigned to adsorbed
crotonaldehyde, since the crotonaldehyde C=O stretch frequency is also typically lowered
upon adsorption.8 The crotonaldehyde vapor-phase exhibits C=O stretch at ~1691 cm-1, as
shown in Table S4.1. The appearance of the former can then be related to acetaldehyde as
well, as the CH2 stretch was also observed by Raskó et al. at 2960 cm-1 on TiO2 and
Ordomsky et al. at 2950 cm-1 over SiO2, ZrO2/SiO2, and MgO/SiO2.8,54 Aldehyde presence
can typically be spotted by its unique carbonyl CH stretch at ~2700 cm-1, however this
peak does not exist in our spectra. The absence of this peak is due to the distorted
acetaldehyde -CHO group, where the C-O bond is now elongated, and the band will shift
to a higher wavenumber at ca. 2800 cm-1, as explained later by DFT results. Ordomsky et
al. had this same observation where the carbonyl CH stretch was only present on the spectra
when there is vapor-phase acetaldehyde on the surface.8 Upon increasing the temperature,
two peaks at 1604 and 1581 cm-1 start gaining intensity; these speaks can be assigned to a
C=C group. We have previously shown that acetaldehyde transformation to surface enolate
has very low activation energy on an Mg3c surface site, which becomes the basis of our
assignment of one of these C=C stretching bands to surface enolate.17 The gas-phase keto-
enol tautomerization of acetaldehyde thermodynamically favors the keto (aldehyde)
form.60 However, Palagin et al. spectroscopically observed proton transfer from
123
acetaldehyde to SnBEA zeolite and attributed this to the surface enolate formation. Further
complicating the assignment of these C=C stretching bands is the rather low Gibbs free
energy barrier of ~16 kcal/mol17 for aldol condensation, which opens up the possibility for
crotonaldehyde formation.
Conflicting assignments have also previously been made for the C=C stretching
bands originating from ethanol adsorption on MgO, where they were also previously
assigned to surface acetate,52,58 and/or 2,4-hexadienal, aldol condensation product of
crotonaldehyde and acetaldehyde.8 The latter was actually shown to form on basic surface
sites at close to dry ice temperatures.61,62 The formation of surface acetate was previously
postulated to take place through an acyl intermediate, which yields a vibrational band at
~1690 cm-1.59 There are several reasons why acyl intermediate is highly unlikely to be
formed on the basic surface: the base-induced Cannizzaro reaction suggests that the
aldehyde needs to be lacking hydrogen in the α-position63, and hence, the only way
Figure 4.6. In-situ DRIFTS spectra acquired of ethanol on WK (1:1) catalyst in 3200
to 1000 cm-1 spectral region. Ethanol was adsorbed on the sample surface and
temperature increased from 373 to 723 K while spectra being recorded. In-situ DRIFTS
spectra of the sample surface with no adsorbate present at every corresponding
temperature were used for reference. In-situ DRIFTS spectra acquired for ethanol
adsorbed on MgO are shown in the inset for 1200 to 1000 cm-1 spectral region.
MgO
124
acetaldehyde can form an acetate is from the strong Lewis acid-driven Tischenko
reaction.64 From the broad shape of the peak it can be concluded that surface acetaldehyde
undergoes enol transformation to yield surface enolates, which then either polymerize
and/or undergo aldol condensation to yield higher aldehydes and bulky aromatic structures.
Collectively, these data show that in addition to physisorption, ethanol chemisorbs
as two different surface species by displacing carbonate structure and producing water as
a byproduct of its adsorption. In addition to the dissociative chemisorption, ethanol adsorbs
as a semi-dissociated species, where the removed proton is still interacting strongly with
the ethoxide, as shown by the frustrated -OH vibration. Part of this structure further
undergo dehydrogenation, which is shown by the acetaldehyde C=O peak, and further
reacts to make C=C containing molecule(s).
Acetaldehyde is often cited as an important intermediate during the catalytic
ethanol transformation to 1,3-BD and has been proposed to have a major role in the overall
reactive mechanism.1,6,23 To further explore the relevant surface intermediates chemistry
and corroborate the assignment of the ethanol IR data, we carried out IR temperature
programmed desorption experiments with acetaldehyde as the probe molecule.
Acetaldehyde IR spectra typically exhibit peaks at 1716, 1680 and 1590 cm-1 which were
related to vapor-phase C=O, adsorbed C=O, and C=C stretches, respectively.65 Other
bands stemming from acetaldehyde adsorption on bifunctional metal oxide surfaces
previously observed were located in the 1450-1350 cm-1 region with conflicting
assignments to either surface acetates66 or, more conservatively, to bending modes of CH,
CH2, and CH3.65,67 Acetaldehyde adsorption was studied over MgO based catalysts, where
on Ni/MgO, it was claimed to undergo several reactions, such as a Cannizzaro reaction
125
which yielded surface acetate peaks around ~1580 and ~1425 cm-1 68 aldol condensation,
as previously observed on MgO/SiO2 by IR spectroscopy.8
Acetaldehyde adsorption on WK (1:1) is shown in Figure 4.7 and tabulated in
Table 4.3. In the C-H region, several bands were observed at 3027, 2967, 2935, 2882,
2845, and 2743 cm-1 upon acetaldehyde introduction to the surface. The first three bands
can be directly assigned to the stretching mode of CH3, while CH3 Fermi resonances can
be assigned to 2882 and 2845 cm-1, These depend on the adsorption site.59 A possible
assignment can also be made that one belongs to acetaldehyde, while the other one to
crotonaldehyde. The weak band at 2743 cm-1 is the signature stretching mode of aldehyde
-CHO group with broadening due to possible crotonaldehyde formation at the low
temperature. The peaks at 1716, 1680, 1633, 1600 and 1578 cm-1 are all related to C=O
and C=C stretches, depending on the binding site and nature of the adsorption/perturbation
of the C=O group. Acetaldehyde undergoes aldol condensation17 and polymerization61,62
with very low energy barriers on basic surfaces, including MgO. The reactivity of
acetaldehyde on the basic surface is the main reason for its rapid deactivation when
acetaldehyde is present in large concentrations.8,69 Peaks at 1456, 1434, 1382, 1346 and
1284 cm-1 can be used to confirm the existence of an acetaldehyde molecule and its reaction
products, such as polymerized acetaldehyde and possible enolate isomerization-aldol
condensation products on the surface. The first two of those peaks, along with the one at
1346 cm-1 , are assigned to the wagging mode of CH3, the mode that polymerized aromatic
acetaldehyde is lacking.59 The peak at 1382 cm-1 and 1284 cm-1 can be assigned to the
bending mode of the acetaldehyde CH3 group59 and wagging mode of the acetaldehyde
CHO group, respectively. However, these bending modes can overlap with those
126
originating from surface enolate, as a result of keto-enol tautomerization, and further
detailed assignments will be made on DFT calculations. There were no peaks observed
below 1300 cm-1 due to the strong SiO2 vibrations and pure MgO data was used to
investigate that spectral region, as shown in the Figure 4.7 inset. In this region, three bands
are observed at 1264, 1107, and 1066 cm-1 when acetaldehyde is observed on MgO. These
bands were previously assigned to η (C-O), η (C-C), and ν (C-C), respectively.59
Analysis of the IR data of both ethanol and acetaldehyde demonstrates that the
surface ethoxy species react to make adsorbed surface acetaldehyde, that, in turn, further
desorb, polymerize, couple, or isomerize as shown by the presence of the peaks at ~1600
and ~1580 cm-1, where intensity in the ethanol spectra is much less than that of
acetaldehyde. This lack of intensity is due to the fact that ethanol would also desorb from
the surface and the steep activation energy, i.e. 39.6 kcal/mol for the dehydrogenation
Figure 4.7. In-situ DRIFTS spectra acquired of acetaldehyde on WK (1:1) catalyst in
3200 to 1000 cm-1 spectral region. Acetaldehyde was adsorbed on the sample surface
and temperature increased from 373 to 723 K while spectra being recorded. In-situ
DRIFTS spectra of the sample surface with no adsorbate present at every corresponding
temperature were used for reference. In-situ DRIFTS spectra acquired for acetaldehyde
adsorbed on MgO are shown in the inset for 1200 to 1000 cm-1 spectral region.
MgO
127
reaction.17 The acetaldehyde formed during the experiments has higher affinity to the basic
surface, which is confirmed by the intense CH stretching and CH3 wagging modes of the
molecule being on the surface even at 723 K. Ordomsky, et al. also acknowledged the
reactivity of the basic catalyst, which, in turn, results in the strong adsorption of
acetaldehyde and/or its higher self-reaction products on the surface.8
2.5.2. C4 intermediates.
DRIFT spectra of crotonaldehyde adsorbed on the WK (1:1) surface are shown in
Figure 4.8. Notably, peaks observed in the spectra are identical with the ones found for
acetaldehyde in Figure 4.7 and tabulated in Table 4.3, except for the relative intensities of
several peaks, such as 1716, 1680, 1650, 1456, and 1434 cm-1. The similarities between
the two spectra suggest potential overlaps between peaks from both aldehydes during the
Figure 4.8. In-situ DRIFTS spectra acquired of crotonaldehyde on WK (1:1) catalyst in
3200 to 1000 cm-1 spectral region. Crotonaldehyde was adsorbed on the sample surface
and temperature increased from 373 to 723 K while spectra being recorded. In-situ
DRIFTS spectra of the sample surface with no adsorbate present at every corresponding
temperature were used for reference.
128
infrared analysis of the surface reactive intermediates. The almost identical spectra make
a solid case for the deduction of acetaldehyde spontaneous coupling to crotonaldehyde via
aldol condensation at 373 K followed by dehydration, which can also be confirmed by the
shoulder at ~1633 cm-1 in Figure 4.7. The interpretation of aldol condensation at 373 K
can be justified by our supplementary TPSR experiments (not shown), which showed that
crotonaldehyde (m/z=70) appeared right when the surface became saturated with
acetaldehyde. The spontaneous reaction of acetaldehyde aldol condensation to yield
hydroxy-butanal and crotonaldehyde had also been previously observed over HZSM-5,65
TiO2, CeO2, and Al2O3.54,58,59 However, infrared spectra alone are not capable of
decoupling both intermediates.
The adsorption of crotyl alcohol on the WK (1:1) surface, as shown in Figure 4.9
and tabulated in Table 4.3, provides more insight for the C=C vibration region. Our data
Figure 4.9. In-situ DRIFTS spectra acquired of crotyl alcohol on WK (1:1) catalyst in
3200 to 1000 cm-1 spectral region. Crotyl alcohol was adsorbed on the sample surface
and temperature increased from 376 to 723 K while spectra being recorded. In-situ
DRIFTS spectra of the sample surface with no adsorbate present at every corresponding
temperature were used for reference.
129
show relatively few peaks and demonstrate the selective nature of the crotyl alcohol
transformation on this bifunctional catalyst. Peaks at 3017, 2965, 2949 and 2840 cm-1 were
assigned to the CH3 and CH2 stretches of adsorbed crotyl alcohol, while peaks at 2932,
2867 and 2738 cm-1 were assigned to both and also to vapor-phase crotyl alcohol.70
Bending and wagging modes of the CH groups can also be seen in the 1400-1300 cm-1
region, where peaks at 1456 and 1441 cm-1 were assigned to ρw (CH2) and ρw (CH3) and
1380 and 1368 cm-1 - to δ (CH2) and δ (CH3). A unique feature of the adsorption is shown
by the presence of an intense, sharp peak at 1602 cm-1 upon adsorption, which red-shifted
~70 cm-1 from the vapor-phase crotyl alcohol peak at 1675 cm-1.70 This peak is assigned to
the C=C stretch of adsorbed 1,3-BD. This was also observed by Wenig and Schrader during
their crotyl alcohol in-situ IR experiments over V-P-O catalyst.71 The accompanying water
peak at ~1650 cm-1 can be seen as a shoulder for the intense 1600 cm-1 peak. Our DFT
calculations discussed later also show that at this adsorbed state, there was no higher CH
peak observed at ~3100 cm-1, due to the elongated C=C group of the C4 structure.
Following a temperature increase to 473 K, a shoulder appeared at 1620 cm-1 and this was
assigned to 1-butene, which also was observed over V-P-O catalyst.71 Alternatively, the
peak at 1602 cm-1 can be assigned to the C=C mode of crotyl alcohol, while the peak at
1620 cm-1 could originate from the adsorbed 1,3-BD, as shown by Cavani et al.10,19 In one
of their experiments, however, this peak appeared when the surface temperature was
increased under inert flow, which, based on our experiments, will not give any higher C4
compounds in the vapor-phase. The peak at 1675 cm-1 is low enough in intensity that it
would be overwhelmed if adsorbed aldehydes were present on the surface. However,
Figure 4.6 shows that there was no peak was found at that wavenumber. This is possibly
130
due to the depleted ethanol and/or the little bit of crotonaldehyde that was being made on
the surface. This argument is supported by the absence of any peaks at ~1700 cm-1,
indicating that there was not enough acetaldehyde on the surface to induce desorption and
aldol condensation. Our observation shows that the formation of acetaldehyde, being very
reactive to the WK (1:1) surface, leads to the formation of various C=C containing
intermediates, including surface enolate, polymerized acetaldehyde, and aldol
condensation products, i.e. crotonaldehyde and 2,4-hexadienal. The crotonaldehyde, does
not, however, proceed further to produce crotyl alcohol and 1,3-BD. The unsaturated
aldehyde tends to remain on the surface of the catalyst, which is why it is rarely observed
in the gas-phase.
2.6. DFT calculations ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol
vibrational frequencies.
Disagreements in peak assignments arise mainly when interpreting the
acetaldehyde adsorption spectra, where polymerization on the catalyst surface takes place
at relatively low temperature,61,62 in addition to the desired aldol condensation8,59,65,72 and
Cannizzaro reaction.66,68 In ethanol adsorption spectra, an important peak at 1604 cm-1,
accompanied by a shoulder at 1581 cm-1, also sparks discrepancy in assignments.8,52,68 Peak
shifts are often observed, mainly in the C-O region, where the anchoring sites vary for
different catalysts and hence almost directly affect the vibration.13,14,30,53 We used DFT
structure optimized and calculated frequencies of both gas and adsorbed ethanol,
acetaldehyde, crotyl alcohol and crotonaldehyde molecules to aid in in-situ DRIFT spectra
assignments. In the calculation, a defect site on MgO (100), i.e. Mg3C2+O4C
2- (kink), was
used for adsorbate structural optimization. Another potential active site, the OH group on
131
MgO, was considered since indirect correlation using XPS has pointed out that more OH
groups on the MgO leads to a better conversion.12 Our detailed analysis using in-situ
DRIFTS, however, showed that the OH groups’ involvement during the reaction is minimal
with its participations limited to substitutional chemisorption and reversible thermal
rearrangement. It is widely known that MgO-SiO2 has defect sites, kinks in particular, that
are stable up to the reaction condition,15,30,31,44 not to mention the recently synthesized MgO
catalyst that gives conversion as good as the SiO2-based material, which questions the
silica’s role on the reaction.12 The use of defect sites has been utilized before10,11,17 and
kink site was chosen as the active site based on the stability comparison that was carried
out by Chieregato et al.10 and Zhang et al.11 It is well known that vibrational frequencies
calculated using harmonic approximation are typically larger than the fundamental ones
observed experimentally with the various scaling factors typically used.73 We calculated
these scaling factors by optimizing ethanol, acetaldehyde, crotonaldehyde, and crotyl
alcohol molecules and comparing the frequencies with the experimental values.70 Using
the method of least-squares we determined the scaling factors to be 0.997, 0.9962, 0.9996,
and 0.9903 for ethanol, acetaldehyde, crotonaldehyde, and crotyl alcohol, respectively. The
scaling factors for all molecules are very close to unity and unscaled frequencies were used
to assign FTIR peaks. The complete scaling calculation can be found in the Supplementary
Information Table S4.1.
We performed frequency calculations for several permutations of ethanol,
acetaldehyde (enolate), crotonaldehyde, and crotyl alcohol adsorbed on a defected MgO
surface.17 The optimized structures are shown in Figure 4.10 and the calculated
frequencies are tabulated in Table 4.4. The two ethanol (I) species lead to acetaldehyde on
132
Mg3CO4C and ethylene on Mg3CO5C, while there are three adsorption configurations of
acetaldehyde (II): physisorbed on Mg3C, chemisorbed on Mg3CO4C and enolate adsorbed
on Mg3CO4C. For crotonaldehyde, configuration (IV) due to the assumed facile dehydration
of acetaldol to crotonaldehyde was used.17 The crotyl alcohol (V) DFT calculation leads to
both the dissociated state and the coordinated 1,3-BD (VI) with the α-C still bound to the
O atom. Structure numbers, as shown in Figure 4.1, refer to the particular steps in the
ethanol catalytic transformation cycle. Acetaldol adsorption was not optimized in this
work due to its rare observation during experiments.1
DFT calculations show that ethanol can be adsorbed in two separate ways: via
dissociative adsorption and semi-dissociative adsorption, with the deprotonated ethoxy still
interacting strongly with the resulting surface hydroxyl. Typical adsorption spectra of
ethanol on an MgO surface show a peak at ~2700 cm-1 which was never previously
discussed.10,14,19 With the periodic DFT calculation, this peak can be assigned to the
stretching mode of the surface hydroxyl group. In Figure 4.6, vapor-phase acetaldehyde
did not appear in infrared spectra. Instead, a peak at 1653 cm-1 appeared at the intermediate
temperatures, and this can be correlated with the DFT-calculated C-O vibration at 1657
cm-1. This is accompanied by the build-up of C=C containing surface species, shown by
the peak at ~1600 cm-1 in Figure 4.6.
Acetaldehyde adsorption on the basic catalyst surface results in the infrared peaks
similar to those of crotonaldehyde (Figure 4.7 and 6). The experimental peak at 1382 cm-
1 was previously assigned to the -CH3 wagging mode of acetaldehyde and this may overlap
with the surface crotonaldehyde, owing to the similarities between Figure 4.7 and 6.59 The
complexity of this peak is demonstrated by the shoulders around it and its appearance in
133
both Figure 4.7 and 6. To answer this question, we looked at crotonaldehyde frequency
calculation as well as the two acetaldehyde species and surface enolate (Figure 4.10). DFT
shows that four of the species all exhibited vibration around that wavenumber, which can
be seen in Table 4.4. Peak at 1284 cm-1 was also previously assigned to the C-O vibration
from acetaldehyde, and this also most likely overlaps with a vibrational mode from
physisorbed acetaldehyde.59 The peak at 1284 cm-1 is revealed to originate from -CHO
bending of an interacting surface species, in which the C=O bond is opened with the
molecule bridging two Mg atoms and one O atom, as shown in Figure 4.10-II. This
vibrational mode is not present on the physisorbed acetaldehyde, where the analog of that
vibration is presented as a peak at 1382 cm-1, contradicting assignment by Singh et al.59
Analysis of the C=C region inevitably presents the possibility that surface enolate is the
more reactive state of acetaldehyde. This enolate is the main building block for further
reactions, such as polymerization and aldol condensation.74 The presence of the enolate is
rather hard to confirm experimentally,74 due to its subsequent spontaneous reactions, but
DFT calculation reveals its presence, as verified the by C=C stretches at 1621 cm-1 and the
-CHO bending mode at 1384 cm-1. Finally, the peak at 2845 cm-1 in Figure 4.8 can be
assigned to the surface crotonaldehyde, blue-shifted from the vapor-phase. Our DFT
calculations show that this peak can be assigned to ν (CH), calculated at 2868 cm-1.
134
I (Mg3CO4C) I (Mg3CO5C) Surface model
II (Mg3C) II (Mg3CO4C) II (Mg3CMg4C)
IV (Mg3CMg4C) V (Mg3CMg4C) VI (Mg3CMg4C)
Mg O
O4C Mg3C υCO = 1133 cm-1
υCH = 2914 cm-1
υCO = 1658 cm-1
υOH = 2678 cm-1
υCO = 1052 cm-1
υCC = 1098 cm-1
υCO = 1068 cm-1
υCC = 1621 cm-1
υCO = 1173 cm-1
υCC = 1661 cm-1
υCO = 1584 cm-1
υCC = 1663 cm-1
υCO = 1081 cm-1
υCC = 1587 cm-1
Figure 4.10. PBE optimized structures of ethanol (I), acetaldehyde (II), its enolate
conformation (II), crotonaldehyde (IV), crotyl alcohol (V) and 1,3-butadiene (VI) on MgO
surface low coordination Mg3cO4c or Mg3cO5c surface sites. Numbers refer to the particular
steps in catalytic transformation cycle shown in Figure 4.1.
135
Table 4.4 Calculated infrared frequencies of ethanol, acetaldehyde, crotonaldehyde and
crotyl alcohol molecules adsorbed on low coordination model MgO surface sites.
Frequencies were calculated using PBE density functional and no scaling to correct for
anharmonicity was applied.
Assignment Ethanol Acetaldehyde Enolate Crotonaldehyde Crotyl
alcohol
Configuration Mg3CO4C Mg3CO5C Mg3C Mg3CO4C Mg3CMg4C Mg3CMg4C Mg3CMg4C
ν (OH) 3574 2679 - - 3063 - -
ν (CH3)
3030,
3020,
2950
3036,
3024,
2952
3096,
3019,
2965
3061,
3037,
2961
- 3000, 2943
3062,
3007,
2959
ν (CH2) 2903,
2873
2938,
2909 - -
3173,
3073 -
2946,
2875
ν (CH) - - 2914 2828 3028 3102, 3068,
3041, 2868
3056,
3040
ν (C=O) - - 1657 1016 1173 1664 -
ν (C=C) - - - - 1621 1581, 1010 1663
δ (CH2) 1460 1469 - - 1384 - 1378
δ (CH3) 1438,
1435
1449,
1440 - - - 1431, 1428 1350
ρw (CH) 1353,
1337 - - 1367 - 1232 -
ρw (CH2) - 1359 - - - - 1434
ρw (CH3) - 1341
1408,
1402,
1320
1447,
1428,
1320
- 1349 1441,
1419
ρw (CHO) - - 1382 1292 1302 1365 -
ρt (CH2) 1260 1267 - - - - -
δ (OH) - 1116 - - 1004 - -
δ (CC) - 1072 1117 1066 - - -
ν (CO) 1133 1052 - - - - -
ρw (CCO) 1132 - - - - - 1081
ν (CC) 1064 - - 1098 - - -
136
The adsorption of crotyl alcohol provides much clarification for the entire reaction
sequence. Silica possesses weak Lewis acid sites and might provide an additional
dehydration site for the reaction. However, when ethanol was run on a bare silica catalyst,
only a little conversion of ethanol was achieved at higher temperature.75 The reaction
yielded acetaldehyde, as well as ethylene and ether, which explains silica’s role as a solid
acid catalyst. MgO has the ability to dehydrate the crotyl alcohol to give 1,3-butadiene,
which justifies the use of MgO defect sites for this reaction.1,75 The interaction with silica,
as shown in the Section 3.3, shows that the strong Lewis acid sites are due to the interaction
of silica and MgO.1,15 The peak at 1600 cm-1, which immediately forms on the surface
during experiment, indicates the presence of a C=C stretch. The vapor-phase crotyl alcohol
exhibits a vibration at around ~1670 cm-1,70 which is blue-shifted ~10 cm-1 for its adsorbed
state, based on the DFT calculation and shown in Table 4. These peaks are also observed
on the spectra in Figure 4.9, with the shifted peak being a shoulder to the intense, sharp
1602 cm-1. This indicates the presence of another surface species, which was shown by
DFT to be a coordinated 1,3-BD (Figure 4.10, VI). The assignment of this 1600 cm-1 peak
is also corroborated by the absence of the sp2 carbon C-H stretch peak, since α-C atom is
still in transition from sp3 to sp2.
2.7. In-situ DRIFT spectra for the ethanol, acetaldehyde, crotonaldehyde and
crotyl alcohol reaction on a WK (1:1) catalyst surface: the effect of the vapor phase
presence.
Formation of the C4 intermediates and products on MgO catalysts requires the
presence of the vapor phase ethanol and does not proceed via adsorbed ethoxide
intermediate catalytic conversion alone.19 This is also supported by our experiments, as
137
discussed in Section 3.5, where ethanol desorption did not lead to the peaks caused by
acetaldehyde on the surface. Hence, we performed in-situ temperature programmed
DRIFTS experiments, during which a continuous reactant flow was carried out over a
sample containing adsorbed intermediates. DRIFT spectra for ethanol adsorbed on WK
(1:1) in the presence of a continuous vapor flow are shown in Figure 4.11. A spectrum of
the catalyst surface with the adsorbate at 373 K was used as a reference. The C-H stretching
region has peaks at 2978, 2933, 2903, and 2877 cm-1. These peaks are all attributed to CH3
and CH2 stretches. These peaks are also accompanied by the triplet at 1456, 1391 and 1061
cm-1 previously assigned to δ(CH3), δ(OH) and υ(CO) of vapor-phase ethanol.70,76 Two
other peaks can also be observed at 1630 and 1322 cm-1 in the low temperature 373 to 473
K regime, while higher temperatures result in their disappearance. While the latter can be
assigned to the wagging mode of the adsorbed ethoxy species, the former was previously
assigned to the adsorbed crotyl alcohol.19 It is highly unlikely, however, that crotyl alcohol
was being made at such a low temperature, since no acetaldehyde was observed. Hence,
we assigned the peak at 1630 cm-1 to adsorbed water in accordance with our assignment in
Section 3.5.1. The higher temperature regime, 523 to 723 K, also consistently resulted in
the greater 1575 and 1440 cm-1 peaks, in addition to peak broadening at ~3061 cm-1 which
appears to be stable on the catalyst surface even at these higher temperatures. The peak
broadening is indicative of the presence of olefins, i.e. ethylene and 1,3-BD. A notable
increase, shown by the 1743 and 1687 cm-1 peaks at intermediate temperatures, can be
assigned to acetaldehyde, both vapor-phase and chemisorbed, respectively. The C=O
stretch peaks coincide with the emergence of a 3004 cm-1 peak, which is attributable to the
C-H stretch of a sp2 carbon. These two peaks are accompanied by a broad band at 1280
138
cm-1, which is the same with the peak at 1284 cm-1 in Figure 4.7. We have assigned this
peak to ρw of CHO. Right after the appearance of this intermediate, peaks at 1579 and 1434
cm-1 become very apparent, suggesting that these vibrations are from the reaction products
of adsorbed acetaldehyde. These two peaks are also similar to those observed in a similar
region, as shown in Figure 4.7 and Figure 4.8 for acetaldehyde and crotonaldehyde surface
adsorption, respectively. As previously assigned, these two peaks originate from a C=C
stretch (1579 cm-1) and the bending modes of CH2 or CH3 (1434 cm-1).
139
Figure 4.11. In-situ DRIFTS spectra acquired of ethanol on WK (1:1) catalyst. Ethanol
was adsorbed on the sample surface, flown continuously and temperature increased from
376 to 723 K while spectra being recorded. In-situ DRIFTS spectrum of the sample
surface with adsorbed ethanol present at 373 K was used for reference.
140
Complementary vapor-phase composition measurements performed using gas
chromatography (not shown) demonstrated that at a temperature regime above 523 K, 1,3-
BD was made but no crotonaldehyde and crotyl alcohol were observed. This suggests that
Figure 4.12. In-situ DRIFTS spectra acquired of acetaldehyde on WK (1:1) catalyst.
Acetaldehyde was adsorbed on the sample surface, flown continuously and temperature
increased from 376 to 723 K while spectra being recorded. In-situ DRIFTS spectrum of
the sample surface with adsorbed acetaldehyde present at 373 K was used for reference.
141
various C4 intermediates to 1,3-BD tend to either react quickly with the product or remain
strongly adsorbed on the surface and react further, instead of desorbing. The correlation
made between the increase in 1,3-BD production by GC and the intensity increase in 1575
and 1440 cm-1 peak intensities implies that these two peaks are predominantly due to a C4
intermediate, with some contributions from surface enolate and polymerized acetaldehyde,
as also previously shown in Section 3.5.1. While the first peak was previously assigned to
2,4-hexadienal,8 surface acetates from the Cannizzaro reaction would exhibit both bands
due to their asymmetric and symmetric -COO stretching modes.52 Their increasing
intensity also suggests that the surface ethoxide, which dehydrogenates to acetaldehyde, is
continuously replenished by the vapor-phase ethanol, as shown in Figure 4.11. This
suggests that ethanol undergoes a catalytic transformation into acetaldehyde on WK (1:1),
which can be regarded as a rate limiting step because the other important intermediates are
spontaneously formed once acetaldehyde is produced. The subsequent aldol condensation
proceeds rapidly even at relatively low temperatures, as shown in Figures 4.5 and 4.6,
while crotyl alcohol is readily dehydrated, as demonstrated by the formation of 1,3-BD at
373 K.17,71 This observation agrees with literature reports, where on a basic catalyst with
little redox properties ethanol dehydrogenation is regarded as the rate limiting step.23,77
In-situ DRIFT spectra of acetaldehyde adsorbed on WK (1:1) in the presence of
vapor are shown in Figure 4.12. Peaks at 3060, 3023, 3000, 2962, 2931, 2870, 2821, 2791,
2736, and 2700 cm-1 are readily observed on these spectra. These various CH3, CH2, CH
sp2 stretches indicate the presence of multiple surface species. As discussed in Sections
3.5.1 and 3.5.2 this is due to the spontaneous polymerization, aldol condensation, and keto-
enol tautomerization of the acetaldehyde. A gradual increase in C-H stretching vibration
142
from 2700 to 2821 cm-1 at lower temperatures is followed by their transformation into the
species responsible for the peaks at 2870 to 2962 cm-1 and can be associated with
transformation of vapor-phase acetaldehyde into chemisorbed acetaldehyde and into
surface enolate and crotonaldehyde. This is different from the temperature programmed
desorption experiments shown in Figure 4.7 and suggests that vapor-phase acetaldehyde
needs to be continuously supplied to replenish the surface species in order to continuously
produce C4 molecules. This experimental observation is also supported by the recent
computational study where aldol condensation of acetaldehyde on an MgO Mg3c site was
shown to result from the interaction between surface enolate and physisorbed
acetaldehyde.17 Peak assignments provided in Table 4.4 also suggest that peaks at 1762,
1724, 1442, 1343, and 1113 cm-1 can be assigned to the presence of vapor-phase
acetaldehyde. Higher temperature regimes above 573 K result in the decrease to 1724 cm-
1 as well as an enhanced 1273 cm-1 peak signifying the conversion of adsorbed
acetaldehyde into acetaldol as an intermediate that intermittently appears before being
dissociated to crotonaldehyde and water. The peak at 1273 cm-1 was previously observed
by Singh et al. and assigned to δ (C-OH) of the aldol.59 The peak at 1616 cm-1 is indicative
of the C=C stretch that originates from enolate, crotonaldehyde or 2,4-hexadienal. The peak
at 1650 cm-1 gradually increased with temperature, indicating the presence of adsorbed
aldehyde which could belong to acetaldehyde or crotonaldehyde. The triplet peak at ~1750
cm-1 slowly transforms into a singlet at higher temperatures indicating the depleted vapor-
phase acetaldehyde, due to the aldol condensation, confirmed by the presence of the
crotonaldehyde peak at 1700 cm-1. Similar to the case of ethanol, peaks at ~1574 and 1442
cm-1 increased with temperature. These peaks are, however, accompanied by a more
143
prominent peak at 1555 cm-1. This peak has not been observed previously in Figures 4.5,
4.6 and 4.9. This ~20 cm-1 red shift is most likely due to the interaction between the C=C
molecule with vapor-phase acetaldehyde. In both ethanol and acetaldehyde reactive
desorption experiments in Figures 4.9 and 4.10 the same native hydroxyl groups at 3747,
3725, and 3680 cm-1 from Table 4.2 and Figure 4.3 are transiently involved in the catalytic
transformations. Furthermore, the hydroxyl group region appears to be similar to those in
Figure 4.5 suggesting no new basic sites are formed or they are immediately consumed by
the ensuing reactions. Figure 4.12 also suggests that aldol condensation on a WK (1:1)
surface proceeds quickly and is not a rate limiting step.
Figure 4.13. In-situ DRIFTS spectra acquired of crotonaldehyde on WK (1:1) catalyst
under ethanol vapor flow. Crotonaldehyde was adsorbed on the sample surface, flushed
with inert gas and ethanol was introduced under continuous flow with temperature
increased from 376 to 723 K while spectra being recorded. In-situ DRIFTS spectrum
of the sample surface with adsorbed crotonaldehyde at 373 K was used for reference.
For comparison, 523 K spectrum of crotonaldehyde adsorbed with no gas phase present
is shown in red dotted line.
144
Meenvein-Ponndorf-Verley (MPV) reduction is a hydrogenation process in which
alcohols are used as a source of hydrogen.78 It was postulated to take place in the reaction
mechanism with ethanol hydrogenating the produced crotonaldehyde.1,16 It is typically
initiated by abstraction of an H+ from the alcohol. It was suggested that the rate limiting
step is hydride transfer from the adsorbed alcohol to the adsorbed carbonyl compound.78
Figure 4.13 shows the corresponding infrared spectra of the adsorbed crotonaldehyde in
the presence of ethanol vapor. Peaks at 2984, 2955, and 2902 can be specifically assigned
to ethanol CH3 and CH2 stretches with some minor contribution by other C4 molecules.
Negative peaks at 2845 and 2743 cm-1, on the other hand, are due to the CH3 Fermi
resonance and ν (CH) of crotonaldehyde, respectively.54,59 Each of these two negative
peaks is accompanied by a positive shoulder at a lower wavenumber, which also increases
with temperature, indicating the presence of another aldehyde, most likely acetaldehyde.
The presence of the vapor-phase acetaldehyde as the side product of the MPV reduction
can be confirmed by the C=O stretch at 1767 cm-1 , which has also been observed in
previous experiments under constant acetaldehyde and ethanol vapor flow. This
confirmation is shown in Figures 4.9 and 4.10. The peak at 1825 cm-1 can’t be assigned
to any of the alcohols or aldehydes. That peak can be found in the gas-phase 1,3-BD IR
spectrum, as reported in the NIST database.70 The signature C=C stretch of 1,3-BD is,
however, impossible to observe due to its overlap with the negative peaks from
crotonaldehyde. Negative peaks can readily be observed in the 1700-1600 cm-1 region and
1520-1400 cm-1 region, indicating the consumption of crotonaldehyde. A red-dotted IR
spectrum is shown as a comparison for the crotonaldehyde desorption at 523 K.
Comparison of the two spectra at the same temperature shows that the intensity decrease
145
of the red plot is much less significant than when ethanol is constantly flown during the
experiment. The depletion of the surface crotonaldehyde is due to temperature-induced
desorption and reduction by ethanol to a certain extent. This intensity decrease is, however,
not accompanied by the peak at 1587 cm-1. This peak is relatively unaffected by the
temperature-programmed reaction, even though there are sharp peaks at 1560 cm-1, which
can also be found in the red-dotted spectrum. The absence of both positive and negative
peaks at 1587 cm-1 indicates that this peak is not part of the reactive intermediate and due
mostly to the overreaction of crotonaldehyde and acetaldehyde to 2,4-hexadienal.8 The
peak at 1520 cm-1 signifies the presence of a C-C containing molecule, which is being
consumed. Interestingly, peaks around 1460 and 1430 cm-1 are both initially consumed
before they start increasing positively. We expect this due to the possible overlap between
several CH3 containing molecules, such as crotonaldehye, which is initially consumed. The
acetaldehyde that is being produced and the vapor-phase ethanol, which is initially
consumed, starts to increase in intensity due to the depleted crotonaldehyde. The intensity
decrease in these peaks is also accompanied by the increasing intensity of the ethanol bands
at 1061, 1286, and 1346 cm-1.
3. Conclusions
Surface chemistry of WK (1:1) catalyst during the reaction of ethanol and the
corresponding reactive intermediates, including acetaldehyde, crotonaldehyde, crotyl
alcohol, has been investigated using in situ DRIFTS measurements combined with DFT
calculations. The nature of the native hydroxyl groups and their reactivity was also
investigated. They were found to undergo a transient reactivity via hydrogen bonded
interactions with the reactive intermediates. Ethanol adsorption resulted in several
146
physisorbed and chemisorbed surface species. Acetaldehyde exhibited high reactivity to
yield crotonaldehyde but the excess resulted in strongly bound surface species assigned to
surface acetate, and/or 2,4-hexadienal or polymerized acetaldehyde. Crotonaldehyde is
more likely to be reduced by ethanol to yield crotyl alcohol than desorbing, even at
relatively high temperatures. Crotyl alcohol, on the other hand, showed to be very reactive
and adsorbs as two different species: physisorbed and deprotonated species that would
further desorb as 1,3-BD. Presence of gas phase hydrogen containing molecules, such as
ethanol, proved to be key in several reactive steps, including acetaldehyde condensation
step and crotonaldehyde reduction. Altogether, the data presented unraveled a complex
interplay between the surface hydroxyl groups, gaseous reactants and surface bound
reactive intermediates of 1,3-BD formation. These complex surface processes are depicted
in Figure 4.14. This elucidated surface reaction mechanism, combined with vapor-phase
intermediate characterization, can be used as a foundation for structure-activity relationship
study in combination with active sites determination. This will further lead to rational
design of catalyst. Future work will attempt to correlate vapor phase product evolution with
the most stable or transient reactive surface intermediates to examine trends leading to
higher overall 1,3-BD selectivity.
147
(I)
(II)
(III) (IV)
(V)
(VI)
Figure 4.14. Complete surface reaction scheme on ethanol reaction over MgO/SiO2
catalyst. (I) Crotonaldehyde, (II) adsorbed crotyl alcohol, (III) 1,3-butadiene, (IV)
2,4-hexadienal, (V) paraldehyde, (VI) metaldehyde.
148
Chapter 4 – Supporting Information
Figure S4.1. In-situ spectroscopy of ethanol on MgO catalyst. Ethanol was adsorbed
on the sample surface and temperature ramped up from 373 to 723 K while spectra
being recorded. Subtracted spectra are shown. Spectra are offset for clarity.
Figure S4.2. In-situ spectroscopy of acetaldehyde on MgO catalyst. Acetaldehyde was
adsorbed on the sample surface and temperature ramped up from 373 to 723 K while
spectra are being recorded. Spectra are offset for clarity.
149
Table S4.1. Calculated infrared frequencies of gas phase ethanol, acetaldehyde, crotyl alcohol and crotonaldehyde molecules.
Frequencies were calculated using PBE density functional and no scaling to correct for anharmonicity was applied. Experimental
frequencies, except for crotonaldehyde, were obtained from NIST.68
Vibration
Ethanol Acetaldehyde Crotyl alcohol Crotonaldehyde
DFT Frequency
(cm-1)
IR
(cm-1)
DFT Frequency
(cm-1)
IR
(cm-1)
DFT Frequency
(cm-1)
IR
(cm-1)
DFT Frequency
(cm-1) IR (cm-1)
ν (OH) 3718 3686 - - 3702 3665 - -
ν CH3 3053, 3038,
2966
3035,
3012,
2960
3092, 3025, 2967
3024,
2996,
2967
3008, 2959 2970 3068, 3011,
2965, 2984
ν CH2 3011, 2925 3008,
2905 - - 3000, 2928
2940,
2880 - -
ν CH - - 2790 2840 3062, 3055, 3037 3030 3095, 3065,
3042, 2788
3044,
3007,
2750,
2828
ν (C=O) - - 1749 1743 - - 1691 1693
ν (C=C) - - - - 1675 1675 1644 1645
δ (CH2,
CH3) 1463 1456 1413, 1406
1410,
1390
1448, 1439,
1425, 1359
1480,
1435,
1415,
1338
1430, 1420,
1353
1449,
1398,
1381
δ (OH) 1325 1391 - - - - - -
δ (COH) - - - - - - 1367 Not
observed
149
150
δ (CCH) 1237 1242 - - - - 1288 Not
observed
Combinati
on
bending
- - - -
1362, 1313,
1285, 1261,
1170, 1115, 1024
1384,
1290,
1250,
1180
1236 Not
observed
Combinati
on stretch - - - - 1075 1080 1142, 1086
1151,
1082
ν (CO) 1027 - - - 981 970 - -
ν C-C
(C=C) 863 - 1095 1122 - - - -
Scaling
factor 0.997 0.9962 0.9903 0.9996
150
151
References
(1) Pomalaza, G.; Capron, M.; Ordomsky, V.; Dumeignil, F. Catalysts 2016, 6 (12),
203.
(2) Shylesh, S.; Gokhale, A. A.; Scown, C. D.; Kim, D.; Ho, C. R.; Bell, A. T.
ChemSusChem 2016, 9 (12), 1462–1472.
(3) Ostromislenskiy, J. J. Russ. Phys. Chem. Soc. 1915, No. 47, 1472 –1506.
(4) Lebedev, S. V. Zh Obs. Khim 1933, No. 3, 698–717.
(5) Posada, J. A.; Patel, A. D.; Roes, A.; Blok, K.; Faaij, A. P. C.; Patel, M. K.
Bioresour. Technol. 2013, 135, 490–499.
(6) Angelici, C.; Weckhuysen, B. M.; Bruijnincx, P. C. a. ChemSusChem 2013, 6 (9),
1595–1614.
(7) Janssens, W.; Makshina, E. V.; Vanelderen, P.; De Clippel, F.; Houthoofd, K.;
Kerkhofs, S.; Martens, J. A.; Jacobs, P. A.; Sels, B. F. ChemSusChem 2014, 8 (6),
994–1008.
(8) Ordomsky, V. V.; Sushkevich, V. L.; Ivanova, I. I. J. Mol. Catal. A Chem. 2010,
333 (1), 85–93.
(9) Muller, P.; Burt, S. P.; Love, A. M.; McDermott, W. P.; Wolf, P.; Hermans, I. ACS
Catal. 2016, 6 (10), 6823–6832.
(10) Chieregato, A.; Velasquez Ochoa, J.; Bandinelli, C.; Fornasari, G.; Cavani, F.;
Mella, M. ChemSusChem 2014, 8 (2), 377–388.
(11) Zhang, M.; Gao, M.; Chen, J.; Yu, Y. RSC Adv. 2015, 5 (33), 25959–25966.
(12) Hayashi, Y.; Akiyama, S.; Miyaji, A.; Sekiguchi, Y.; Sakamoto, Y.; Shiga, A.;
Koyama, T.; Motokura, K.; Baba, T. Phys. Chem. Chem. Phys. 2016, 18 (36),
25191–25209.
(13) Hanspal, S.; Young, Z. D.; Shou, H.; Davis, R. J. ACS Catal. 2015, 1737–1746.
(14) Birky, T. W.; Kozlowski, J. T.; Davis, R. J. J. Catal. 2013, 298 (0), 130–137.
(15) Kvisle, S.; Aguero, A.; Sneeden, R. P. A. Appl. Catal. 1988, 43 (1), 117–131.
(16) Toussaint, W. J.; Dunn, J. T.; Jachson, D. R. Ind. Eng. Chem. 1947, 39 (2), 120–
125.
(17) Taifan, W. E.; Bučko, T.; Baltrusaitis, J. J. Catal. 2017, 346, 78–91.
(18) Quattlebaum, W. M.; Toussaint, W. J.; Dunn, J. T. J. Am. Chem. Soc. 1947, 69,
593–599.
(19) Velasquez Ochoa, J.; Bandinelli, C.; Vozniuk, O.; Chieregato, A.; Malmusi, A.;
Recchi, C.; Cavani, F. Green Chem. 2016, 18, 1653–1663.
(20) Velasquez Ochoa, J.; Malmusi, A.; Recchi, C.; Cavani, F.; Ochoa, J. V.; Malmusi,
A.; Recchi, C.; Cavani, F. ChemCatChem 2017, 9 (12), 2128–2135.
(21) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A.
ChemSusChem 2014, 7 (9), 2505–2515.
(22) Makshina, E. V.; Janssens, W.; Sels, B. F.; Jacobs, P. A. Catal. Today 2012, 198
(1), 338–344.
(23) Makshina, E. V; Dusselier, M.; Janssens, W.; Degreve, J.; Jacobs, P. A.; Sels, B. F.
Chem. Soc. Rev. 2014, 43 (22), 7917–7953.
(24) Wu, C.-M.; Baltrusaitis, J.; Gillan, E. G.; Grassian, V. H. J. Phys. Chem. C 2011,
115 (20), 10164–10172.
(25) Haas, P.; Tran, F.; Blaha, P. Phys. Rev. B 2009, 79 (8), 85104.
152
(26) Kantorovich, L. N.; Holender, J. M.; Gillan, M. J. Surf. Sci. 1995, 343 (3), 221–
239.
(27) Chizallet, C.; Petitjean, H.; Costentin, G.; Lauron-Pernot, H.; Maquet, J.;
Bonhomme, C.; Che, M. J. Catal. 2009, 268 (1), 175–179.
(28) Chizallet, C.; Bailly, M. L.; Costentin, G.; Lauron-Pernot, H.; Krafft, J. M.; Bazin,
P.; Saussey, J.; Che, M. Catal. Today 2006, 116 (2), 196–205.
(29) Di Valentin, C.; Del Vitto, A.; Pacchioni, G.; Abbet, S.; Wörz, A. S.; Judai, K.;
Heiz, U. J. Phys. Chem. B 2002, 106 (46), 11961–11969.
(30) Branda, M. M.; Rodríguez, A. H.; Belelli, P. G.; Castellani, N. J. Surf. Sci. 2009,
603 (8), 1093–1098.
(31) Bailly, M.; Chizallet, C.; Costentin, G.; Krafft, J.; Lauronpernot, H.; Che, M. J.
Catal. 2005, 235 (2), 413–422.
(32) Baltrusaitis, J.; Hatch, C.; Orlando, R. J. Phys. Chem. A 2012, 116 (30), 7950–
7958.
(33) Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49 (20), 14251–14269.
(34) Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6 (1), 15–50.
(35) Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54 (16), 11169–11186.
(36) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 48 (17), 13115–13118.
(37) Blöchl, P. E. Phys. Rev. B 1994, 50 (24), 17953–17979.
(38) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77 (18), 3865–3868.
(39) Knözinger, E.; Jacob, K.-H.; Singh, S.; Hofmann, P. Surf. Sci. 1993, 290 (3), 388–
402.
(40) Coluccia, S.; Lavagnino, S.; Marchese, L. Mater. Chem. Phys. 1988, 18 (5), 445–
464.
(41) Anderson, P. J.; Horlock, R. F.; Oliver, J. F. Trans. Faraday Soc. 1965, 61 (0),
2754–2762.
(42) Shido, T.; Asakura, K.; Iwasawa, Y. J. Chem. Soc.{,} Faraday Trans. 1 1989, 85
(2), 441–453.
(43) Morrow, B. A. Stud. Surf. Sci. Catal. 1990, 57, A161–A224.
(44) Chizallet, C.; Costentin, G.; Che, M.; Delbecq, F.; Sautet, P.; Surface, D.; Marie,
P. J. Am. Chem. Soc. 2007, No. 19, 6442–6452.
(45) Hair, M. L. J. Non. Cryst. Solids 1975, 19, 299–309.
(46) Chung, S.-H.; Angelici, C.; Hinterding, S. O. M.; Weingarth, M.; Baldus, M.;
Houben, K.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS Catal. 2016, 6 (6),
4034–4045.
(47) Hofmeister, A. M.; Bowey, J. E. Mon. Not. R. Astron. Soc. 2006, 367 (2), 577–
591.
(48) Mellini, M. Am. Mineral. 1982, 67 (5–6), 587–598.
(49) Taifan, W.; Boily, J.-F.; Baltrusaitis, J. Surf. Sci. Rep. 2016, 71 (4), 595–671.
(50) Vimont, A.; Thibault-Starzyk, F.; Daturi, M. Chem. Soc. Rev. 2010, 39 (12), 4928–
4950.
(51) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A. Catal.
Sci. Technol. 2015.
(52) Song, H.; Ozkan, U. S. J. Catal. 2009, 261 (1), 66–74.
(53) Phung, T. K.; Lagazzo, A.; Rivero Crespo, M. Á.; Sánchez Escribano, V.; Busca,
G. J. Catal. 2014, 311, 102–113.
153
(54) Raskó, J.; Kiss, J. Appl. Catal. A Gen. 2005, 287 (2), 252–260.
(55) Busca, G.; Lorenzelli, V. Mater. Chem. 1982, 7 (1), 89–126.
(56) Wang, H.; Schneider, W. F. Phys. Chem. Chem. Phys. 2010, 12 (24), 6367–6374.
(57) Meixner, D. L.; Arthur, D. A.; George, S. M. Surf. Sci. 1992, 261 (1), 141–154.
(58) Raskó, J.; Dömök, M.; Baán, K.; Erdőhelyi, A. Appl. Catal. A Gen. 2006, 299,
202–211.
(59) Singh, M.; Zhou, N.; Paul, D. K.; Klabunde, K. J. J. Catal. 2008, 260 (2), 371–
379.
(60) Cederstav, A. K.; Novak, B. M. J. Am. Chem. Soc. 1994, 116 (9), 4073–4074.
(61) Georgieff, K. K. J. Appl. Polym. Sci. 1966, 10 (9), 1305–1313.
(62) Bevington, J. C.; Norrish, R. G. W. Proc. R. Soc. London. Ser. A. Math. Phys. Sci.
1949, 196 (1046), 363 LP-378.
(63) Cannizzaro, S. Justus Liebigs Ann. Chem. 1853, 88 (1), 129–130.
(64) Tishchenko, V. E. J. Russ. Physico-Chemical Soc. 1906, 38 (355–418).
(65) Diaz, C. D. C.; Locatelli, S.; Gonzo, E. E. Zeolites 1992, 12, 851–857.
(66) Idriss, H.; Diagne, C.; Hindermann, J. P.; Kiennemann, A.; Barteau, M. A. J.
Catal. 1995, 155 (2), 219–237.
(67) Guil, J. M.; Homs, N.; Llorca, J.; De La Piscina, P. R. J. Phys. Chem. B 2005, 109
(21), 10813–10819.
(68) Resini, C.; Cavallaro, S.; Frusteri, F.; Freni, S.; Busca, G. React. Kinet. Catal. Lett.
2007, 90 (1), 117–126.
(69) Ho, C. R.; Shylesh, S.; Bell, A. T. ACS Catal. 2016, 6 (2), 939–948.
(70) Stein, S. E. In NIST Chemistry WebBook, NIST Standard Reference Database
Number 69; Linstrom, P. J., Mallard, W. G., Eds.; National Institute of Standards
and Technology (NIST): Gaithersburg, MD, 2017.
(71) Wenig, R. W.; Schrader, G. L. J. Phys. Chem. 1987, 91 (22), 5674–5680.
(72) Rekoske, J. E.; Barteau, M. a. Langmuir 1999, 15 (6), 2061–2070.
(73) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100 (41), 16502–16513.
(74) Palagin, D.; Sushkevich, V. L.; Ivanova, I. I. J. Phys. Chem. C 2016, 120 (41),
23566–23575.
(75) Sekiguchi, Y.; Akiyama, S.; Urakawa, W.; Koyama, T.; Miyaji, A.; Motokura, K.;
Baba, T. Catal. Commun. 2015, 68, 20–24.
(76) Plyler, E. K. J. Res. Natl. Bur. Stand. (1934). 1952, 48 (4).
(77) Niiyama, H.; Morii, S.; Echigoya, E. Bull. Chem. Soc. Jpn. 1972, 45 (3), 655–659.
(78) Hattori, H. Chem. Rev. 1995, 95 (3), 537–558.
(79) Towns, J.; Cockerill, T.; Dahan, M.; Foster, I.; Gaither, K.; Grimshaw, A.;
Hazlewood, V.; Lathrop, S.; Lifka, D.; Peterson, G. D.; Roskies, R.; Scott, J. R.;
Wilkins-Diehr, N. Comput. Sci. Eng. 2014, 16 (5).
154
Chapter 5
Active Sites Determination of MgO/SiO2
Catalysts for Ethanol to 1,3-BD Reaction
Abstract .......................................................................................................................... 154
1. Introduction ....................................................................................................... 155
2. Results and Discussion ...................................................................................... 160
2.1. Steady state ethanol catalytic conversion to 1,3-BD ................................. 160
2.2. Bulk, surface chemical and structural characterization using XRD, LEIS
and DRIFTS ........................................................................................................ 163
2.3. Temperature-programmed reaction spectroscopy (TPRS) of ethanol on
MgSi-WK ............................................................................................................ 167
2.4. Acid-base characterization using DRIFTS ................................................ 172
2.5. Reactive site persistence during ethanol-to-1,3-BD .................................. 174
2.6. Implications for the structure-activity relationship .................................... 186
3. Conclusions ........................................................................................................ 189
References ...................................................................................................................... 191
Abstract
Ethanol is an important renewable chemical that allows for sustainable high value
product, such as 1,3-butadiene, catalytic synthesis. MgO/SiO2 catalyst is typically utilized
in a single step ethanol-to-1,3-butadiene catalytic conversion and the (by)product yields
were shown to depend on the type, structure and strength of the catalytic active sites. The
fundamental factors describing the molecular structure and binding properties of these sites
is thus of critical importance but not yet fully understood. We utilized multimodal
approach, including temperature programmed surface sensitive infrared, mass
155
spectroscopy using probe molecules, such as CO2, NH3, pyridine and propionic acid, to
unravel the structure and persistence of these catalytic sites in situ. In particular, Mg-O-
Mg, Mg-O(H)-Mg, Mg-O-Si and Mg-O(H)-Si surface site binding configurations were
interrogated using spectroscopic methods in combination with DFT calculations. Surface
elemental analysis using low energy ions suggested that either Mg atoms or Si being the
most abundant on the topmost surface layer, depending on the catalyst preparation method.
The molecular active site structure was determined and incipient wetness prepared surface
was found to be dominated with stabilized Mg-OH with little magnesium silicate (Mg-O-
Si and Mg-O(H)-Si) functional groups. The wet-kneaded catalyst surface, on the other
hand, contained a significant number of surface sites derived from magnesium silicates.
The fundamental surface site structure proposed here can further serve as a starting point
for theoretical calculations necessary to fully model the reactive pathway during ethanol
catalytic transformation to 1,3-butadiene.
1. Introduction
Elucidating surface active site structure is of high importance for development of
selective MgO/SiO2 catalysts utilized for the catalytic conversion of ethanol to 1,3-
butadiene (1,3-BD). Deemed as Lebedev catalyst, it is increasingly investigated due to its
bifunctional nature and the variety of the (by)products that can be formed thereon.1–4 The
nature of the closed shell MgO electronic structure and diverse surface functionality
provides interesting challenges as there are much fewer spectroscopic methods that allow
identification of the active site properties, akin to those of solid Lewis acid catalysts, such
156
as Ta2O55 or ZrO2,
6 also used in ethanol-to-1,3-butadiene catalytic conversion (1,3-BD).3,7
As a result, the surface site structure and the reactivity of MgO/SiO2 catalysts during 1,3-
BD production are still poorly understood.8 Furthermore, discrepancies between the
selectivity of the reported active catalysts are due to the intrinsic active site density, their
functional nature (acidic or basic) and their strength which arise from the diverse set of
preparation methods including ratio of Mg-to-Si, Mg precursor used and their deposition
method. The ratio of acidic-to-basic sites was shown to affect the overall reactivity during
1,3-BD formation, as demonstrated by Angelici et al.9 Shylesh et al. proposed that weak
basic sites were responsible for ethanol dehydrogenation and other basic sites for aldol
condensation.3 Angelici et al. attributed higher overall reactivity to a small number of
strong basic sites on the catalyst surface with an intermediate amount of acidic sites and
weak basic sites.9 Catalysts prepared using different methods, e.g. incipient wetness
impregnation (IWI) and wet-kneading (WK)3,9 resulted in a large activity difference with
IWI prepared catalyst yielding only ~5% conversion at 300 °C when compared to WK
catalyst ~50% conversion at 425 °C. The key to this different activity was believed to be
the significant improvement of acetaldehyde production by transition metal promotion on
strong basic sites.3 In general, MgO/SiO2 wet-kneading has consistently been shown to
produce highest 1,3-BD yields4,10,11 due to the proposed balanced acidic and basic catalytic
site number.9 The exact molecular structure of these acidic and basic active sites is still
under debate with most analysis focused correlating the reactivity and the (bulk) crystalline
catalyst phases.1,2,12 In particular, -Mg-O-Si- linkage has been implicated to be reactive and
related to the selectivity of the catalysts.1–3 SiO2 was proposed to indirectly catalyze the
reaction due to its structural perturbation of MgO using wet-kneading.10 Furthermore, a
157
solid solution of MgO with SiO2 was inferred from the experimental measurements and
the amount of magnesium silicate phases, measured by 1H-29Si cross-polarized MAS
NMR1,3 and DRIFTS,2 was correlated to varying overall selectivity. Ochoa et al. observed
formation of a magnesium silicate phase with Mg2+ neighbored by Si4+ cations synthesized
using sol-gel methods with Mg/Si ratio of 9 to 15 while the lower ratio led to the formation
of catalytically inactive forsterite (Mg2SiO4) phase formation.2 Amorphous magnesium
silicate hydrous phase formed during wet-kneading of MgO and SiO2 was found to be
responsible for ethylene byproduct formation while layered magnesium silicate hydrous
phase was correlated to the 1,3-BD product.1 Furthermore, the silicate-to-MgO ratio was
suggested to be the key to the appropriate balance of acidic-basic sites.1 A different view
was offered by Shylesh et al. where hydroxyl (OH) groups were necessary in the proximity
of the strong basic Mg2+-O2- sites to synergistically catalyze the reaction. Finally, the
correlation between the magnesium silicate hydrous phase and 1,3-BD yield was
challenged by Hayashi et al. who reported MgO catalyst that did not require participation
from SiO2 for this reaction.12 Said MgO catalyst was synthesized with an additional
hydrothermal step using NH4OH solution. XPS characterization of the two different MgO
catalysts, i.e. with and without the additional hydrothermal step, showed that latter, i.e. the
more active catalyst, exhibited a higher intensity of an unassigned O1s oxygen peak at
around 532 eV.12 The presence of this unidentified oxygen species on MgO could be related
to the reactive lower-coordinated oxygen atoms on Mg-O defect sites.13–15 Concurrently,
these lower-coordinated Mg-O pairs (Mg2+3CO2-
4C, Mg2+3CO2-
3C, Mg2+4CO2-
4C) were
computationally shown to be involved in 1,3-BD formation from ethanol.16–18 Analysis of
Lewis acid - ZrO2-based catalysts19 - suggests that Lewis acid sites (LAS) can be chiefly
158
responsible for the activity in this reaction. By definition, closed Lewis acid heteroatoms
(M) are tetrahedrally coordinated (M-(OSi)4 to the zeolite framework, while open Lewis
acid heteroatoms are tri-coordinated (HO)-M-(OSi)3 to the zeolite framework.20–22 With
this in mind, the octahedral symmetry in MgO crystal12,23 allows to identify several LAS
as part of the intrinsic acid/base pairs to be available. i.e. Mg-O-Mg, Mg-O(H)-Mg, Mg-
O-Si and Mg-O(H)-Si. These combinations can further exist in open and closed acid
configurations, where the oxygen is bound to SiO2 while also coordinated to a proton to
form coordinated hydroxyl groups.24,25 Strict terminology of the open acid site requires an
isolated hydroxyl group to be present and while it is very basic, this hydroxyl group
spectroscopically has been proven to be non-existent.25,26 In addition to these sites, the
coordination of Mg is also very important, since catalysis by this metal oxide is driven by
defect sites.15,27,28 These proposed catalytic sites are shown in Figure 5.1.
While Zr-based catalysts mainly concerns the Zr-coordination into the framework,
and consequently, characterization of the resulting LAS, i.e. open and closed,19,29 study on
MgO/SiO2-based catalysts mostly revolves around the general acidity and basicity
characterization while also discussing the importance of bulk silicate phases.1,9,30,31
However, pyridine-DRIFTS studies concluded LAS to be the only acidic sites on the
catalyst as demonstrated by the IR peaks at 1450, 1578, and 1612 cm-1.3,9,30,31 NH3-TPD,
another routinely used acidity probe method, also discriminates the acid based on the
strength, without discussing the nature of the acidic sites.9 Hence the molecular structure
of these acidic sites is not well known.1 In this work, we combined spectroscopic
measurements in-situ using different probe molecules to identify the role of each sites
during the reaction and elucidate their molecular coordination. In particular, we begin by
159
performing bulk XRD, surface LEIS and DRIFTS analysis of native surface hydroxyl
groups. We then perform steady state and kinetic temperature programmed experiments
of ethanol conversion to 1,3-BD using catalysts synthesized with different methods. We
then utilize temperature programmed DRIFTS to explore surface acidic and basic site
structure with ab initio calculations to support our NH3 adsorption site assignments and
hence propose the molecular arrangements of the catalytic sites. Sodium (Na) poisoning
was utilized to elucidate the role of the acidic sites during the reaction, which will further
indicate the importance of strong acidic sites during the reaction. Finally, the persistence
of these reactive sites is probed spectroscopically under the relevant conditions of
temperature and ethanol vapor.
B
C
D
E F
A
Figure 5.1. Possible combination of metal atoms that act as Lewis acid sites: A: Mg3C
(open), B: Mg3C (closed), C: Mg4C (closed), D: Mg4C (open), E: Mg5C (open), F: Mg5C
(closed).
160
2. Results and Discussion
2.1. Steady state ethanol catalytic conversion to 1,3-BD
Steady state reactivity of the synthesized MgO/SiO2 catalysts was investigated
using fixed-bed reactor. A catalyst calcined at 500°C was synthesized as a benchmark
catalyst. This benchmark calcination temperature was based on earlier study by Zhu, et al.,
where 500°C was the optimized calcination temperature that resulted in a balanced acidic-
basic sites at 40.8 µmol/g v 49 µmol/g.32 Activity comparison was performed at a
temperature of 450 °C with the maximum 1,3-BD yield. At this reaction temperature,
carbon balance for each catalyst was determined to be >80%. Table 5.1 shows that ethylene
selectivity was the highest for MgSi-WK2 which suggests the presence of acidic sites on
the surface since ethanol dehydration reaction is very prominent over catalysts with very
high density of BAS.33–35 For catalysts calcined at 800 °C (MgSi-WK and MgSi-IWI)
ethylene selectivity was above 50%, which is intriguing, since pyridine probing does not
show the presence of BAS (vide infra). This ethylene formation can be proposed due to the
reactive Mg-O-Mg or Mg-O-Si linkages that are inherent in the catalysts.36 In agreement,
DFT calculations have shown that ethanol dehydration competes with dehydrogenation
reaction over LAS in MgO catalysts.16
Comparison between two catalysts calcined at 800 oC, i.e. MgSi-WK and MgSi-
IWI, shows that MgSi-WK was more active and selective to 1,3-BD suggesting that the
preparation method deeply affected the balance of the active sites. The accumulation of
acetaldehyde on MgSi-IWI catalyst was evident from Table 1 suggesting the active sites
for further aldol condensation and MPV (Meerwein-Ponndorf-Verley) reduction –
mechanistic steps taking place after ethanol dehydrogenation - were limited.16 MgSi-WK,
161
on the other hand, exhibited significantly higher 1,3-BD selectivity and limited
acetaldehyde production suggesting more sites available for the subsequent reactions.
Figure 5.2 shows the evolution of each major (by)product with increasing temperature for
reaction over MgSi-WK catalyst. 1,3-BD and ethylene exhibited almost linear increase in
productivity with the calculated apparent activation energy of 12.4 and 18.02 kcal/mol,
respectively. Importantly, Arrhenius plot shown in Figure 5.2 inset of acetaldehyde
formation did not show linearity due to its involvement into further reactions.
Table 5.1 Steady state reactivity of MgO/SiO2 catalysts of different calcination
temperature and preparation method. Reaction was carried out at 450 °C with catalyst
mass of 0.1 g, 55 ml/min total flow rate and pethanol = 2.5 kPa. Selectivity towards major
(by)products ethylene, acetaldehyde and 1,3-BD is reported.
Catalyst Selectivity (%) Conversion
(%) Ethylene Acetaldehyde 1,3-BD
MgSi-WK 55.8 14.4 29.7 77.0
MgSi-WK2 82.9 9.7 7.5 60.4
MgSi-IWI 58.0 25.6 16.4 63.9
Productivity values of MgSi-WK compared reasonably well with those found in the
literature. In the present work 1,3-BD yield translated to the production rate of 0.44 gBD.gcat-
1.hr-1, while the MgSi-WK2 yielded about 0.06 gBD.gcat
-1.hr-1. Chung et al. synthesized a
WK catalyst using calcination temperature of 500 °C with the reported 1,3-BD productivity
of 0.23 gBD.gcat-1.hr-1.1 The origin of this reactivity can be related to the high amount of
layered hydrous magnesium silicate phase on the catalyst, which was highly dependent on
the precursor; a nano-sized Mg(OH)2 precursor was the preferred precursor.1 The effect of
MgO precursor used during the WK catalyst synthesis has recently been highlighted by
Huang et al. where Mg(OH)2 precursor synthesized using a template method yielded a very
high productivity of 1.15 gBD.gcat-1.hr-1.4 Since precursor was not the controlled variable in
162
the present work, the origin of higher activity in the case of MgSi-WK, as compared to
MgSi-WK2, is presumably from the higher calcination temperature. The effect of the
calcination temperature has been observed previously for Zr/SiO2 catalysts where highest
1,3-BD yield was obtained with catalyst calcined at 550 °C.37 Study on the effect of
calcination temperature on MgO/SiO2 catalysts, on the other hand, revealed linear increase
in basicity with increasing calcination temperature.38 Pyridine testing indicated that
calcination temperature of 500 °C resulted in a catalyst that exhibited the highest acidity.38
Incipient wetness impregnation (IWI) method is not typically used for the MgO/SiO2
catalysts synthesis for this reaction despite its popularity in supported catalyst synthesis.
Typically, IWI method is utilized to obtain sub-monolayer coverage to prevent the
formation of second layer of bulk oxides. The activity of this catalyst was very different
yielding a much lower 1,3-BD productivity. Shylesh et al. investigated a similar catalyst
calcined at 500 °C and reported rather low 1,3-BD yield of 0.01 gBD.gcat-1.hr-1 at 300 °C.3
Surface layer compositions of this catalyst as well as MgSi-WK will further be interrogated
using LEIS to better understand chemical composition changes leading to such different
reactivity.
Importantly, steady state experiments suggest that ethylene and acetaldehyde are
the most encountered byproducts during the reaction. As shown in Table 5.1, their
combined selectivity makes up to more than 70% of the total activity. While acetaldehyde
recycling can be utilized due to it being a reactive intermediate, ethylene production should
be limited. One of the advantages of using MgO/SiO2 is no butene byproduct production.
Butenes form azeotrope with 1,3-BD and increase any separation cost. It was reported that
butenes could be formed by hydrogenation of 1,3-BD over platinum catalysts, dehydration
163
of n-butanol, and from thermal or catalytic dimerization of ethylene.39 Hence n-butanol
dehydration might not be feasible considering the high selectivity of ethylene. However,
although the reactive pathway is similar, n-butanol synthesis requires a tautomerization site
to convert crotyl alcohol to 1-buten-1-ol over a basic oxygen site, which further requires
another MPV reduction site.40 This pathway is not supported by our catalytic, DRIFTS36
and TPRS data (vide infra) since no n-butanol was formed in the product stream and no
butyraldehyde was spectroscopically observed.
2.2. Bulk, surface chemical and structural characterization using XRD, LEIS and
DRIFTS
Bulk crystalline structure of the catalysts was characterized using XRD. XRD
patterns of the selected MgSi-WK catalysts as a function of the corresponding oxide ratio
as well as those for MgSi-IWI are compared in Figure 5.3. XRD pattern of WK catalysts
indicates the formation of periclase MgO in the bulk. The intensity expectedly enhanced
Figure 5.2. Catalytic activity of MgSi-WK between 350-450°C. Inset: Arrhenius plot
of ethylene and 1,3-BD. Catalyst mass = 0.1 gr, total flow rate = 55 ml/min, pethanol =
2.5 kPa.
164
with higher Mg content which means that the catalyst crystallinity originated from MgO
rather than any silicate material. On the other hand, IWI catalyst showed a very different
crystalline structure. Several crystalline phases were identified in the XRD patterns
including those that might be attributed to periclase phase of MgO. However, most of the
peaks were due to the presence of different crystalline phase, potentially magnesium
silicates.41–44 While this provided implications for the varied selectivity shown in Table
5.1, surface chemical analysis was performed to further elucidate this effect.
Topmost surface layer of the two catalysts, i.e. MgSi-IWI and MgSi-WK, was
probed using LEIS. LEIS is by far the most surface sensitive characterization technique
which sputters the surface with very low energy ions using ionized noble gases.45 LEIS
spectra of both catalysts are shown in Figure 5.4. Sputtering experiments (depth profiling)
are shown as insets where surface layers were sputtered using 1 keV Ar+ ions. The
sputtering rates were on the order of one monolayer of atoms per 1015 Ar+ ions/cm2. The
legend in the insets indicates the nth layer removed from the catalyst surface. Three peaks
Figure 5.3. Comparison of XRD patterns of MgSi-WK and MgSi-IWI. WK with
different oxide ratios are overlaid for comparison.
165
were found in the spectra corresponding to oxygen at ~1200 eV, Mg at ~1650 eV and a
shoulder for Si at ~1760 eV. The increase in signal intensity for all peaks after initial
sputtering may be due to some residual overlayer on the surface or perhaps to initial
planarization of the catalyst granules increasing the apparent global atomic surface density.
Depth profile spectra were obtained stepwise using the acquisition of the surface spectra
followed by sputtering for a designated period. The resulting depth profiles show Mg-rich
surface for MgSi-IWI while Si-rich surface can be for MgSi-WK. The incipient wetness
impregnation method deposited the magnesium nitrate precursor on top of the fumed SiO2
yielding a high magnesium content. The wet-kneaded method, on the other hand, provided
intimate mixing between the Mg(OH)2 and SiO2 allowing for the extensive interaction
between the two oxides which is reflected in the abundance of Si on the surface.1 Based on
this characterization, MgSi-IWI should contain more Mg-O-Mg linkages, while MgSi-WK
would contain more Mg-O-Si linkages.
The structure of the native OH groups of all the catalysts was investigated using
DRIFTS (Figure 5.5). MgSi-WK shows four major peaks in the spectrum in addition to a
Figure 5.4. Depth-profile of (a) MgSi-IWI and (b) MgSi-WK as probed using HS-LEIS.
HS-LEIS spectra of layer by layer sputtering of catalyst surface are shown in the inset.
166
broad peak at ~3550 cm-1 assigned to germinal and vicinal OH groups of the silica
support.46,47 Peak at 3745 cm-1 was assigned to both isolated silanol group of the SiO2
support46,47 which decreased in intensity when MgO was wet-kneaded and to an isolated
OH group of MgO that depends on the coordination number of the Mg.36 An intense peak
at 3680 cm-1 was previously assigned to a magnesium silicate phase, lizardite.1 The other
peaks at 3725 and 3705 cm-1 were assigned to the isolated O4c-H and O5c-H coordinated
groups formed in the presence of the amorphous SiO2 (SiMg4cO4c and SiMg4cO5c).36
Formation of the peaks at 3725 and 3705 cm-1 was also confirmed with varying Mg/Si ratio
as shown in Figure 5.5 (right). At very high Mg content, i.e. Mg/Si > 7/3, it can be seen
that a very basic Mg-OH group started to become apparent at 3765 cm-1. As previously
shown,36 MgO possessed two basic OH groups that depend on the coordination number of
oxygen atoms with the lower coordinated OH group at higher wavenumbers.25,48
Interestingly, the peaks at 3705 and 3680 cm-1 increased with Mg content while the peak
at 3725 cm-1 intensified at intermediate ratio and diminished at both extremes. This
suggests that the peaks at 3680 and 3705 cm-1 were dominant at lower Si content but not
in pure MgO, and hence, it can be assigned to an OH group anchored to MgSi coordination
that has few Si atoms nearby. The formation of this peak confirms the XRD assignment
where the catalyst is becoming increasingly crystalline MgO-like, both on the surface and
in the bulk.
The OH groups that were observed on MgSi-IWI surface were very different from
those found on MgSi-WK. Two new peaks at 3610 and 3571 cm-1, in addition to the peaks
that were also found on MgSi-WK, appeared, indicating the formation of two entirely new
OH groups. From HS-LEIS experiment, the top surface layer mostly consisted of Mg
167
atoms, which would suggest the presence of a lot of Mg-OH groups. As previously
explained, the isolated OH groups from MgO exhibit peaks at both ~3760 and ~3740 cm-
1, depending on the coordination number. The presence of SiO2 likely stabilized the lowest-
coordinated OH group, i.e. peak at ~3760 cm-1, converting it into the higher-coordinated,
isolated OH group, i.e. peak at ~3740 cm-1. The other peaks at 3610 and 3571 cm-1 are
sharp with lower intensity, unlike the broad OH peaks that were assigned to multi-
coordinated hydrogen-bonded OH groups,24,25 These two peaks also were observed by
Ochoa et al. in the catalysts structurally similar to forsterite.2 From HS-LEIS, XRD, and
DRIFTS, the surface of MgSi-IWI was confirmed to be mainly populated by Mg-OH, with
significantly less magnesium silicate phases. Most of the formed magnesium silicate
phases were found in the bulk of the catalyst as indicated by the XRD.
2.3. Temperature-programmed reaction spectroscopy (TPRS) of ethanol on MgSi-
WK
To further understand the reaction mechanism, temperature-programmed reaction
spectroscopy (TPRS) was carried out using the ethanol reactant and intermediate reactive
Figure 5.5. Left: Comparison of OH groups of MgSi-WK and MgSi-IWI as probed by
in-situ dehydrated DRIFTS experiments; right: OH groups of WK catalysts with
different oxide ratios.
168
molecules. The TPRS was performed on MgSi-WK surface due to the higher reactivity
than MgSi-IWI. Surface chemistry of this catalyst was previously interrogated using in
situ DRIFTS36 with the reaction mechanisms proposed based on the studied surface species
formed during the reaction. TPRS provides vapor-phase analysis allowing to fully
understand the mechanism as a prelude to the analysis of the catalytic sites responsible.
Table 5.2 shows the m/z utilized for reactive vapor (gas) species detection.
Table 5.2 m/z selection to identify the arising vapor-phase species from TPRS experiments
m/z Species
46 ethanol
26 ethylene
44 acetaldehyde
2 hydrogen
54 1,3-BD
70 crotonaldehyde
57 crotyl alcohol
MgSi-WK catalyst was pretreated at 500 °C for 1 hour to simulate the real operating
conditions. The adsorption of ethanol was performed at 100 °C to avoid any residual water
vapor condensation and flushed with inert gas to remove loosely bound molecules.
Experiment was performed under constant ethanol feed flow and shown in Figure 5.6
(left). The ethanol signal continuously decreased during the reaction as a function of
temperature without the presence of any products detected. This is consistent with the
previous report where significant amount of reactive intermediates was bound strongly to
the catalyst suface.36 Ethylene was the first product to be detected at ~200 oC which can be
explained by ethylene’s lower desorption energy than acetaldehyde.16 Acetaldehyde, when
formed, would tend to stay on the surface to undergo several other surface reactions, such
as aldol condensation and polymerization.36 The more significant consumption of ethanol
169
taking place at 300 oC, where acetaldehyde and hydrogen were detected at the same time
around 350°C, which can be explained by the accelerated dehydrogenation reaction. Very
low signal of crotyl alcohol and crotonaldehyde were also evident from the spectra which
indicated the tendency of these species to undergo surface reaction than desorb off the
surface.
Evolution of the reactive intermediates and byproducts during acetaldehyde
temperature programmed experiment is shown in Figure 5.6 (right). When only
acetaldehyde was reacted over the sample crotonaldehyde was formed, in good agreement
with DRIFTS experiments reported previously.36 Several desorption temperature peaks
were observed at 210, 330, 410 °C (acetaldehyde) and 210, 350, 422 °C (crotonaldehyde).
The low temperature peak, i.e. 210 °C was due to the low temperature aldol condensation
between the two acetaldehyde molecules. The presence of ethanol during the experiment
imposes competitive surface MPV reduction and desorption of crotonaldehyde, as
previously suggested36 and in the absence of ethanol led to higher desorption rate of
crotonaldehyde. The presence of crotonaldehyde was supported by the presence of m/z=45.
Figure 5.6. TPRS spectra of ETB reaction over MgSi-WK with ethanol as the feed (left)
and acetaldehyde as the feed (right). EtOH: ethanol; AA: acetaldehyde; CA:
crotonaldehyde; C-OH: crotyl alcohol.
170
This m/z can also be associated to 3-hydroxybutanal (acetaldol). The spectrum looks very
similar to that of crotonaldehyde which indicates a facile dehydration of the acetaldol
formed on the surface during the aldol condensation. Reverse reaction of acetaldol on the
surface was also expected when ethanol was not present on the surface since it would shift
the equilibrium to the left when the resulting crotonaldehyde is not reacted.49 This is
suggested by the relatively lower intensity of the m/z=45 between 300-350 °C than that of
crotonaldehyde in combination with the increasing acetaldehyde signal. The sudden
change in the slope of the TP (temperature-programmed) peak of all discussed m/z, i.e. 44,
45, and 70, indicates an additional different reaction mechanism for aldol condensation.
Palagin et al. suggested an alternative mechanism for aldol condensation without the
enolization step.50 Comparison between H-D exchange experiments of Sn-BEA, Zr-BEA
and Ti-BEA demonstrated that enolized acetaldehyde was only stabilized over Sn-BEA. A
separate mechanism took place where an open Lewis-bound acetaldehyde interacted with
a second acetaldehyde adsorbed on the opposing OH group of the catalyst. DFT calculation
showed that the activation energy of this second mechanism was more than triple than that
of the enolization mechanism (~2 eV v ~0.6 eV).50 A second peak at 330 and 350 °C for
acetaldehyde and crotonaldehyde, respectively, indicated a secondary reaction that takes
place. As previously suggested,36,51 accumulation of acetaldehyde on the catalyst surface
will lead to aldol condensation between crotonaldehyde and acetaldehyde to yield 2,4-
hexadienal. This reaction was confirmed by the change of slope from the water signal
m/z=18 which increased when the other signals decreased.
Finally, TPRS experiments were conducted with both ethanol and acetaldehyde
(Figure 5.7). Acetaldehyde was first preadsorbed on the surface, flushed with an inert gas
171
to remove the physisorbed molecules and temperature ramp was performed under a
constant ethanol flow. This experiment mimics a two-step reactive process where
acetaldehyde is cofed with ethanol. If acetaldehyde production is the rate-determining step
(RDS) any accumulation of acetaldehyde on the surface would increase 1,3-BD production.
Surprisingly, this experiment did not improve production of 1,3-BD as one would expect
ethanol immediately undergo MPV reduction with the produced crotonaldehyde on the
surface. Rather, 1,3-BD production was low until 360 °C which is much later for the
ethanol alone. The sudden increase of the 1,3-BD production was accompanied by water
production suggesting the dehydration of crotyl alcohol was lagging until 360 °C as well.
This production onset coincided with a marked ethanol signal decline suggesting the MPV
reduction becoming the RDS then acetaldehyde/crotonaldehyde is accumulated on the
surface. The increase of acetaldehyde signal was mostly from both activated ethanol
dehydrogenation and MPV reduction byproduct since H2 also increased at higher
temperatures.
Figure 5.7. TPRS spectra of ETB reaction over MgSi-WK with ethanol and
acetaldehyde as the coreactants. Acetaldehyde is pre-adsorbed on the surface and
temperature ramp is under ethanol flow.
172
2.4. Acid-base characterization using DRIFTS
The acidity and basicity of the catalysts was investigated using CO2 and pyridine as
probe molecules (Figure 5.8). DRIFTS was used to qualitatively describe the acid and
basic sites that are present on the surface. CO2 adsorbs on basic surface sites as surface
carbonate and bicarbonate species.52 The formation of these surface species can be
associated with the strength of the corresponding basic sites.34,36 Adsorption of CO2 on
MgSi-WK resulted in three major, broad peaks at 1655, 1541 and 1406 cm-1, while MgSi-
IWI showed broad peaks at 1645, 1620, 1505, 1406 and 1385 cm-1. These peaks were
assigned to carbonates and bicarbonate formation on MgO site since CO2 adsorption on
SiO2 should not yield any surface species.3 In particular, the fundamental doubly
degenerate υ3(COO-) vibration of carbonate on MgO is assigned to bidentate with ~1650
(υ3as) and ~1300 cm-1 (υ3
s) and monodentate carbonate ~1550 (υ3as) and ~1400 cm-1 (υ3
s)
while bicarbonate is detected at ~1650 (υ3as) and ~1380 cm-1 (υ3
s).53,54 These three species
were present on MgSi-WK catalysts, demonstrated by peaks at 1655 and 1325 cm-1
(bidentate carbonate), 1541 and ~1420 cm-1 (monodentate carbonate) and ~1670, 1406,
and 1220 cm-1 (bicarbonate). On MgSi-WK, the dominant peaks were those originating
from monodentate carbonate and bicarbonate. MgSi-IWI, on the other hand, exhibited
rather different surface chemistry. Monodentate carbonate was apparent from the peaks at
1505 and 1385 cm-1, while peaks at 1645 and 1406 cm-1 were assigned to adsorbed
bicarbonate. The presence of bidentate carbonate might be signified by the peak at 1620
and a very small peak at ~1300 cm-1. The strength of the same species on these two catalysts
was very different, as shown by the corresponding DRIFT spectra acquired at 450 °C. At
this reaction temperature, peaks at 1561 and 1368 cm-1 were apparent either due to the
173
surface species rearrangement or due to the new species of lower intensity not detected at
lower temperatures. These two peaks were assigned to monodentate carbonate due to the
narrow υ3 split and stability at higher temperature.34 The stronger basicity of the MgSi-IWI
originated from its Mg-rich surface, as supported by the LEIS spectra.
Pyridine is a weak base and can discriminate between strong and basic site although
its use is limited by its relatively large molecule size, in comparison to NH3.55 Pyridine
adsorbs on a catalyst surface as a physisorbed molecule, Lewis-bonded species and as a
pyridinium ions.55–57 Observation of the latter two species allows to discriminate between
the LAS and Brønsted acid sites (BAS).58 BAS, which would be indicated by a peak at
1638 and 1540 cm-1,58 was not observed which aligns well with the observations of
Angelici et al. and Janssens et al.9,30 Both catalysts exhibited similar Lewis acidity as
shown at 450 °C as evident from peaks at 1445, 1590 and 1605 cm-1. The peaks at 1445
and 1608 cm-1 shifted to 1450 and 1608 cm-1 at higher temperature, i.e. 450 °C, while the
peak at 1490 cm-1, which is a combination band of both LAS and BAS, disappeared. Based
on both HS-LEIS and DRIFTS experiments, the origin of this Lewis acidity should be
different for both catalysts where MgSi-WK would acquire its acidity from unpaired
Figure 5.8. Acid-base characterization of MgSi-IWI and MgSi-WK catalysts probed
using CO2 (left) and pyridine (right). Spectra at high temperature (450°C) and low
temperature (100°C) are shown.
174
electrons of oxygen atom in the Mg-O-Si coordination while MgSi-IWI mostly from the
Mg-O-Mg coordination (vide supra).
2.5. Reactive site persistence during ethanol-to-1,3-BD
Far less explored is reactive site acidity and basicity characterization (and
persistence of the active sites) under reactive conditions of ethanol at reaction temperatures.
To elucidate the role of the specific basic and acidic sites during the reaction on the aged
catalyst, in-situ characterization experiments were performed using CO2 and pyridine after
ethanol reaction on MgSi-WK. In the first experiment ethanol was adsorbed at 100 °C for
20 min, flushed with inert gas for 1 hour and the resulting surface sites were probed with
CO2 or pyridine in-situ. In the second experiment, reaction with ethanol was carried out at
200 °C to initiate ethanol dehydrogenation. After 1 hour of reaction the reaction cell was
flushed and CO2 or pyridine was introduced into the cell. This way nature of the reactive
sites, their persistence and availability for reaction were measured.
Figure 5.9 (left) shows spectra resulting from CO2 adsorption. Three distinct peaks
at 1615, 1380 and 1330 cm-1 appeared on the catalyst that was previously exposed to
Figure 5.9. In-situ acid-base characterization of MgSi-WK catalyst before and after
ethanol adsorption at 100°C and reaction at 200°C using CO2 (left) and pyridine (right).
175
ethanol (MgSi-WK EtOH(ads)) at 100 °C that do not originate from CO2. While peaks at
1380 and 1330 cm-1 were assigned to ρw (CH) and ρw (CH3) of surface ethoxide,36 peak at
1615 cm-1 can be associated with a monodentate carbonate, accompanied by the
symmetrical vibration peak at ~1320 cm-1. After ethanol adsorption all peaks decreased in
intensity, as compared to CO2-only adsorption on the unreacted catalyst at the same
temperature. This significant decrease indicates a competitive adsorption between ethanol
and CO2 that resulted in surface ethoxide and both monodentate and bidentate carbonate.
This indicates that ethanol preferably adsorbed on the strong basic sites that would form
monodentate and bidentate carbonate when exposed to CO2. Reaction at 200 °C was
performed in-situ before degassing with inert and adsorption of CO2. The reaction
temperature was chosen of 200 °C to limit further reactions to 1,3-BD. Extensive inert
degassing was done to limit the further C4 oxygenates formed from facile aldol
condensation, dehydration, and polymerization from occupying the site.36 Comparison of
the spectra between CO2 adsorbed on activated catalyst and aged catalyst, i.e. extensively
reacted at 200°C, showed decrease in intensity on all peaks related to all surface species
arising from CO2. Weak basic sites, which are represented by peaks at ~1650 and ~1400
cm-1 (surface bicarbonate),53 were depleted. This indicates the consumption of the weak
basic sites during the reaction. These weak basic sites on MgO/SiO2, would be the OH
groups, since the strong Mg2+-O2- pairs form monodentate and bidentate carbonate when
exposed to CO2.
Pyridine probing of the acidic sites shows a non-discriminative trend for both
ethanol adsorption at 100 oC and reaction at 200 oC. LAS indeed were consumed during
the adsorption and even more so after the reaction at 200°C, as indicated in Figure 5.9
176
(right). No generation the new of BAS was observed after either adsorption or reaction as
indicated by the absence of peaks at 1545 and 1638 cm-1.58 The Mg-O-Mg linkage present
on MgO exhibits Lewis acidity to a certain extent, with the absence of Brønsted acidity.50
The LAS on MgSi-WK are fundamentally represented by the Mg atoms in four groups:
Mg-O(H)-Mg, Mg-O(H)-Si, Mg-O-Mg, and Mg-O-Si, with the first two groups being the
open LAS. These were further distinguished by coordination number of the magnesium
atoms: the lower the coordination, the stronger the atom is due to the electron deficiency
of the cation. Although pyridine can’t discriminate open LAS from closed sites, its
combination with the conducted CO2 DRIFTS experiment provides more information on
the involved group. The two sites are distinguished by the consumed bicarbonate species,
which is formed when CO2 is adsorbed to a site containing OH group. Hence, the
consumption of both LAS and bicarbonate site (weak basic site), can be traced to the open
LAS. The two open sites, i.e. Mg-O(H)-Mg and Mg-O(H)-Si, were discriminated by the
strength of the base pair. The former is less likely to participate during the reaction, due to
its very basic nature. However, activation of this group has been observed when a bare
MgO is activated using NH3-thermally treated MgO.12 The Mg-O-Si linkages have
previously been correlated to the enhanced activity1,2 while open LAS had been shown to
be responsible for the increased 1,3-BD production.19
To further elucidate the role of acidic and basic sites during the reaction to 1,3-
butadiene, surface site poisoning experiments were carried out using probe molecules such
as CO2, propionic acid and NH3 in a steady state fixed bed reactor. CO2 and propionic acid
are two weak acids while NH3 is a basic probe molecule. Figure 5.10a shows the effect of
cofeeding with CO2. Slow, steady decrease in acetaldehyde and 1,3-BD production before
177
CO2 was introduced indicates slow catalyst deactivation. However, once the CO2 was cofed
to the system, formation rate of 1,3-BD and ethylene dramatically dropped, while
acetaldehyde production increased. CO2 is a weaker acid than propionic acid and will bind
to the strongest basic sites. As shown in Figure 5.8, CO2 did not typically adsorb to the
surface at 450 °C. This poisoning effect suggests that CO2 poisoned the sites that catalyzed
aldol condensation and the subsequent steps, since more acetaldehyde was released into
the vapor-phase without further reacting. When CO2 flow was switched off, the production
of acetaldehyde, 1,3-BD and ethylene was restored, confirming the weak interaction
between CO2 and the strong basic sites.
Figure 5.10b shows (by)product formation rates upon the introduction of propionic
acid. All three products showed a decline in formation rate. When propionic acid
concurrent flow was stopped, the production of acetaldehyde was restored but 1,3-BD and
ethylene formation did not recover. Propionic acid interacted more strongly with the
stronger base sites but also binds to any weaker basic sites. Hence, when propionic acid
flow was stopped, only weak basic sites were accessible while some of the strong basic
sites were permanently poisoned. From the two experiments it is evident that acetaldehyde
production was catalyzed by weak basic sites and 1,3-BD - by strong basic sites. The
production of acetaldehyde over the weak basic site is consistent with the DRIFTS
performed using CO2 as probe molecule (Figure 5.9). Similar phenomenon was observed
by Shylesh, et al. where the 1,3-BD formation rate did not recover during propionic acid
cofeeding experiment over the Au-promoted IWI MgO/SiO2 catalyst.3 Very interestingly,
ethylene formation during the cofeeding experiments followed the trend of 1,3-butadiene.
As previously suggested, ethylene formation can be carried out over both Lewis acidic
178
oxygen atoms (LAS) in the Mg-O-Mg or Mg-O-Si sites and the acidic O-H group
(BAS).16,59 Poisoning with both CO2 and propionic acid affected the strong and medium
Lewis basic Mg atoms in the Mg-O-Mg and Mg-O-Si which inevitably perturbed the strong
Lewis acid pair, i.e. oxygen anions, as well.
Figure 5.10. Acid-base poisoning reactivity testing using (a) CO2, (b) propionic acid,
and (c) NH3 to determine the role of each site during ethanol conversion to 1,3-BD over
WK-800 MgO/SiO2 catalyst. Reactions are carried out at 425 °C, mcat = 0.1 g, pethanol =
2.5 kPa, total flow = 55 ml/min. All formation rates are normalized to initial 1,3-BD
formation rate.
179
NH3 is a relatively strong gas-phase base, stronger than pyridine and other organic
basic molecules, such as acetonitrile and benzenes.55 At reaction temperature of 425 °C,
NH3 exhibits very weak adsorption on the surface and would interact with the acidic sites.
Evident from Figure 5.10c, acetaldehyde production was hardly affected by the poisoning,
as opposed to 1,3-BD and ethylene, where the decrease in production was very pronounced.
While ethylene formation inhibition was reversible, 1,3-BD formation was irreversibly
affected. NH3 poisoned both strong and weak BAS and LAS but when its flow was
discontinued only the strong BAS were poisoned. Ethylene synthesis trend was very similar
for both propionic acid and NH3 cofeeding which indicates the same acid-base pairs being
poisoned during the experiment. 1,3-BD production involves two dehydration steps and
the poisoning indicates that its production did not require participation from the site that
dehydrates ethylene. From these experiments, it is evident that dehydration steps of both
acetaldol and crotyl alcohol were catalyzed by strong acidic sites while ethanol dehydration
was catalyzed by weaker acidic sites.
To confirm participation of the acidic sites, Na2O, a strong basic oxide, was added
to the catalyst in a post-treatment step that permanently poisoned some of the acidic sites.
Three Na concentrations were chosen in this study, i.e. 250, 500, and 1000 ppm, labeled as
250NaMgSi-WK, 500NaMgSi-WK, and 1000NaMgSi-WK. Ethylene productivity was
significantly limited with up to 50% suppression, as shown in Figure 5.11b. Figure 5.11a
shows that sodium poisoning limited the productivity of 1,3-BD as well. The only condition
that gave comparable 1,3-BD productivity to the undoped catalyst was at 450 °C and with
250NaMgSi-WK catalyst. This limitation was also reflected in the acetaldehyde
production. As in 1,3-BD case there was no trend observed with varying Na content but
180
acetaldehyde production over Na2O-poisoned catalysts was higher than for the unpromoted
catalysts. This suggests that Na2O poisoned the acidic sites unselectively eliminating both
BAS and LAS that are responsible for ethylene and 1,3-BD production. The effect of
acidity modification by poisoning with alkali metal was also investigated by Da Ros et al.60
However, the loss of acidic sites in their catalysts was offset by the presence of Zr and Zn
which provided extra Lewis acidity, which was presumably responsible for selectively
dehydrate the C4 molecules to 1,3-BD.
The origin of the poisoning effect was correlated with CO2 and NH3 DRIFTS
experiments. Figure 5.12a shows the appearance of the peaks at 1643, 1463, 1372, and
1356 cm-1 in addition to the native basic sites of the MgSi-WK. Peaks at 1463 and 1356
cm-1 reached a maximum for 500NaMgSi-WK before decreasing in intensity at 1000 ppm,
while the shoulder at 1372 cm-1 became obvious with higher Na loading. The stability of
these peaks indicates the presence of bicarbonates, as shown by the peaks at 1643 and 1372
(1356) cm-1. The peak at 1463 cm-1 is sharp and also is not stable at high temperature which
indicates the presence of another weakly adsorbed CO2 species. Na2CO3, on the other hand,
possesses a band at this specific wavenumber, i.e. 1463 cm-1.61 Na2CO3 is a very stable
carbonate melting prior to decomposing to Na2O, starting at 850°C.62 At higher
temperature range, i.e. 300-450°C an interesting observation emerges from the growth of
the peaks at ~1600 and ~1370 cm-1, as shown in Figure 5.12a (top) for 1000NaMgSi-WK.
The presence of the shoulders to these peaks suggests the presence of two distinct carbonate
groups. From the split, i.e. ~300 cm-1, these peaks are characteristics of bidentate carbonate.
The growth of this peak suggests that rearrangement of (bi)carbonates took place upon
181
thermal treatment. Furthermore, the stability of this peak at high temperature indicates the
enhanced basic site when Na was introduced to the catalyst.
The acidity of the Na-modified catalysts was probed with using NH3 as a probe
molecule, as shown in Figure 5.12b. On pure dehydroxylated MgO, NH3 adsorbs as
physisorbed molecule, with a peak at ~1605 cm-1 while two different LAS exhibited
vibrational peaks at >1605 and 1560 cm-1.63,64 On hydroxylated surface, however, the
interaction is more complicated due to the contributions of OH groups, where hydrogen
Figure 5.11. Productivity of (a) 1,3-BD, (b) ethylene, and (c) acetaldehyde of Na-
poisoned MgSi-WK catalysts between 350-450°C. Catalyst mass = 0.1 gr, total flow rate
= 55 ml/min, pethanol = 2.5 kPa.
182
bonding between NH2H—OH (1612 cm-1) and H3N—HO (1634 cm-1) obscured the
DRIFTS spectra.63,64 The peak at 1430 cm-1 was assigned to ammonium ion (NH4+) as a
result from interaction between ammonia and a BAS.63 Physisorbed ammonia species was
observed on both oxygen and hydroxyl group by the appearance of peaks at 1554 and 1605
cm-1, respectively.63,64 Presence of BAS during NH3 but not pyridine probing had been
observed in the past, where pyridine underrepresented the amount of acidic sites.9,65 This
discrepancy was due to the size of the molecule with NH3 being more mobile.9 The BAS
that was found on the catalysts was isolated, less-accessible and hence might participate
less during the reaction. Four peaks that are assignable to LAS are recognized on these
catalysts by the peaks at 1560, 1580, 1600, 1620, and 1650 cm-1. To aid the assignments
of these peaks DFT vibrational frequency calculations performed on defect sites of MgO,
i.e. Mg3CO4C and Mg4CO4C for both open and closed LAS (Figure 5.13). The corresponding
vibrations were tabulated in Table 5.3.
Table 5.3. Comparison between observed experimental values of NH3 adsorption on MgSi-
WK catalysts with DFT calculated IR vibrations of NH3 adsorbed on open and closed acid
Mg3C and Mg4C sites. Scaling factor of 0.9854 was applied to the calculated values and was
derived from the gas-phase NH3 experimental and DFT calculated frequencies.
Type Mg coordination Vibrational mode Experimental values
δas H-N-H δs H-N-H δas H-N-H δs H-N-H
Open 4C 1592 1566 1600 1560
3C 1574 1534 1570 1540
Closed 4C 1588 1571 N/A N/A
3C 1620 1577 1620 1580
Assuming an ideal surface and similar trend on Mg-O-Si sites, peak assignments
can be readily made. For instance, experimentally, MgSi-WK (0 ppm Na) catalyst
exhibited both closed and open LAS. Closed LAS were evident from the peaks at 1620 and
1580 cm-1 (Mg3C) while open LAS from Mg4C were recognized by peaks at 1600 and 1560
183
cm-1. The presence of peak at 1540 cm-1 signified the presence of another open LAS (Mg3C)
which would be accompanied by a shoulder peak at ~1570 cm-1. The assignments of these
peaks here were made based on the comparison with the DFT computed values tabulated
in Table 5.3. The shoulder at 1650 cm-1 is obvious and can’t be ignored. Echterhoff and
Knözinger attributed a peak at 1634 cm-1 to hydrogen bonding between ammonia and
surface hydroxyl (NH2-H—HO-Mg).64 This vibrational mode is entirely possible due to
the identified NH4+ on the surface indicating the presence of some BAS. Alternatively, if
a Si atom replaces one Mg atom in the Mg3C-O4C-Mg4C (closed LAS) and results in Mg-
O-Si linkage then a change in electronegativity will occur and the magnesium atom
becomes more positively charged which would result in the shorter bond between the Mg
and N atoms. The shorter bond would result in the shift of the peak to a higher wavenumber
in this case higher than 1620 cm-1. Increasing Na loading resulted in the decreasing LAS
and BAS, which was expected. The 250 ppm Na-poisoned catalyst, surprisingly, exhibited
higher intensity of all peaks associated with LAS and BAS, indicating the enhanced acidity
at this temperature. The enhanced acidity, however, was only intermittent at this
temperature, since the top spectra in Figure 5.12b shows the better stability of NH3 surface
species on unpoisoned catalyst at elevated temperature, i.e. 300 °C.
184
Figure 5.12. Bottom: DRIFTS characterization of Na-doped MgSi-WK using (a) CO2
and (b) NH3. Spectra are taken at 100°C after extensive evacuation with N2. Top: (a)
CO2 desorption spectra of 1000 ppm Na-doped MgSi-WK at 100, 300, and 450°C and
(b) NH3 desorption on 0 ppm and 250 ppm Na-doped MgSi-WK at 300°C. Spectral
subtraction was done using the spectra of the dehydrated catalysts at respective
temperatures as the background.
185
(b)
Figure 5.13. (a) MgO periodic model used for DFT simulation of NH3 adsorption on
MgO Lewis acid sites: (b) Mg3C, closed, (c) Mg4C, closed, (d) Mg3C, open, (e) Mg4C,
open. Multiple possible adsorption sites, i.e. kink (Mg3CO4C), edge (Mg4CO4C), and
planar (Mg5CO5C) are highlighted.
(a)
(c)
(d) (e)
Mg3C
Mg3C O4C
O4C
Mg4C
Mg4C
O4C
O4C
Mg4C
O4C
O4C
O4C Mg4C
Mg4C
O5C
Mg3C O4C Mg4
C
O4C
Mg5C O5C
186
2.6. Implications for the structure-activity relationship
The wet-kneading method provides a deeper, more intimate mixing between MgO
and SiO2.1 Incipient wetness impregnation, on the other hand, is a typical synthesis method
for the synthesis of supported catalyst. This method is appropriate when the target catalyst
is a well-dispersed, below monolayer, metal oxide that is supported on a high surface area
support. However, at ratio of 1, the ‘promoter’ is at similar amount with the support, which
corresponds to layered metal oxide on the surface. The nature of the synthesis method
would result in two different bulk phases, i.e. MgO-phase and SiO2-phase, which are
bridged by an interface that should equally contain both oxides. Figure 5.14 schematically
represents these three phases that are formed during the synthesis method. In the case of
wet-kneading, the extensive interaction between MgO and SiO2 allows the boundary phase,
i.e. the middle part in Figure 5.14, to grow larger, as confirmed by the LEIS experiment
Figure 5.14. Schematic diagram to show the presence of various sites investigated with
NH3 and CO2 DRIFTS experiments. The basic sites (orange) are shown in the figure as
both Brønsted base (OH) and Lewis site (electron accepting oxygen atoms), and acid sites
(blue) are represented as Brønsted acid sites (H) and Lewis acid sites (electron donating
magnesium and silicon atoms).
187
and in-situ dehydrated DRIFTS of the catalysts. The MgO-rich phase that is formed on the
catalyst contributes to the higher amount of strong basic sites, which is evident from CO2-
probing comparison, shown in Figure 5.8. The reduced amount of strong basic sites makes
MgSi-WK a better catalyst than MgSi-IWI, as demonstrated by the higher 1,3-BD
formation.
As confirmed by TPRS, the catalytically relevant step during the transformation of
ethanol to 1,3-BD was determined to be acetaldehyde formation step. The ex-situ
characterization alone indicated the importance of weaker basic sites during the reaction
and in-situ poisoning experiment with CO2 and propionic acid confirmed the need for the
weaker basic sites for the reaction. In particular, poisoning the strong basic sites with CO2
resulted in higher acetaldehyde formation rate which suggests that the weak basic sites, not
poisoned by CO2, catalyzed ethanol dehydrogenation and the stronger basic sites catalyze
the subsequent reaction steps. This finding agrees that of Shylesh et al. where propionic
acid was used as a poisoning agent.3 The acidic sites are responsible for both 1,3-BD and
ethylene formation considering the nature of the reaction steps. From NH3 poisoning
experiments it is evident that the weaker acidic sites are responsible for ethanol dehydration
while the stronger sites are responsible for acetaldol and crotyl alcohol dehydration. The
reasoning behind this is the strong interaction between crotyl alcohol, crotonaldehyde and
the surface.36 Poisoning the strong acidic sites resulted in the accumulation of the heavy
aromatic compounds on the surface, which is also confirmed by the unprecedented amount
of aromatic carbonaceous compound observed after the NH3 reaction. The indispensable
role of acidic sites on the reaction is further confirmed by Na-poisoned catalysts, where,
188
although ethylene formation was virtually suppressed, 1,3-BD formation did not
significantly benefit from the reduced acidic sites.
As suggested by the characterization using in-situ DRIFTS and LEIS, the mixing
between MgO and SiO2 would allow the formation of Mg-O-Si linkages which is
consistently observed by previous investigators.1–3 With the possibility of hydroxylation of
the surface, there are four possible formation of the LASs, i.e. open and closed acid sites
of both Mg-O-Mg and Mg-O-Si. The LASs, i.e. Mg2+ cations, will present different
strength, depending on the coordination and type of linkage. Our NH3-DRIFTS experiment
demonstrated the formation of several distinct Lewis sites of different strength. The in-
situ acid-base characterization after ethanol reaction at 200 °C showed that weak basic sites
and LASs were depleted. Schematic representation of this site can be seen in Figure 5.15.
Figure 5.15. Representation of the role of basic sites during ethanol conversion to
acetaldehyde. Top figure represents dehydrated (pretreated) catalyst; bottom figure
demonstrates the absence of bicarbonate when CO2 is adsorbed in-situ after reaction at
200 °C.
189
The weak basic sites discussed are both the Mg-O(H)-Mg and Mg-O(H)-Si. Since at 200
°C only ethylene and acetaldehyde were produced this suggested that the open sites were
the most catalytically relevant active sites. It should be noted, however, that this linkage
contains both acidic and basic sites in the form of Mg2+ cations and OH group, respectively.
Hence, the strength of Mg-O(H)-Mg and Mg-O(H)-Si should be different. The basicity for
the latter would be weaker, as suggested by the in-situ DRIFTS, where the peaks for these
linkages are well below that of Mg-OH, i.e. 3680, 3705, and 3725 cm-1 (Figure 5.5). The
strength of the acidic site, i.e. Mg cation, can be hypothesized to be lower as well in the
case of Mg-O(H)-Si due to the electron density, since Si cations possess atomic charge of
+4e. This leads to the proposed both Mg-O(H)-Si to be the sites that are responsible for
ethanol dehydrogenation and ethylene dehydration while stronger acidic sites and basic
sites are responsible for C4 dehydration and ethanol dehydrogenation. The exact molecular
structure of the various open (closed) acidic and basic sites proposed is shown in Figure
5.14 and includes both LAS and BAS.
3. Conclusions
MgO/SiO2 catalyst active surface sites were analyzed using in situ DRIFTS (using
complementary DFT calculations), TPRS and steady state reactor in combination with bulk
XRD and surface LEIS measurements. Acid-base characterization showed that IWI
synthesis method resulted in a highly basic catalyst with reactive properties originating
from the abundance of Mg atoms on the topmost surface layer, as opposed to WK catalyst.
The molecular active site structure was determined and MgSi-IWI surface was found to be
dominated with stabilized Mg-OH with little magnesium silicate hydroxyl groups. The
190
MgSi-WK surface, on the other hand, contained a significant number of surface sites
derived from magnesium silicates as indicated by the distinct OH groups. This fundamental
site structure difference consequentially led to a different reactivity where MgSi-WK
possessed a more balanced weak-strong basic sites than the basic sites present on MgSi-
IWI. From various reacting molecule poisoning experiments it was determined that the
weak basic sites were responsible for ethanol dehydrogenation, strong basic sites for aldol
condensation and MPV reduction, while stronger acid sites catalyze acetaldol and crotyl
alcohol dehydration reactions and weak acid sites catalyzed the undesired ethanol
dehydration. Furthermore, through a combination of NH3-TPD and DFT the presence of
open and closed LAS was identified while further elaborating Mg coordination, as adopted
from LAS classification of zeolitic materials.20–22 The MgSi-WK catalyst was shown to
have both open LAS with both Mg3C and Mg4C as the anchoring LAS, while also a very
isolated closed LAS (Mg3C).
191
References
(1) Chung, S.-H.; Angelici, C.; Hinterding, S. O. M.; Weingarth, M.; Baldus, M.;
Houben, K.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS Catal. 2016, 6 (6),
4034–4045.
(2) Velasquez Ochoa, J.; Bandinelli, C.; Vozniuk, O.; Chieregato, A.; Malmusi, A.;
Recchi, C.; Cavani, F. Green Chem. 2016, 18, 1653–1663.
(3) Shylesh, S.; Gokhale, A. A.; Scown, C. D.; Kim, D.; Ho, C. R.; Bell, A. T.
ChemSusChem 2016, 9 (12), 1462–1472.
(4) Huang, X.; Men, Y.; Wang, J.; An, W.; Wang, Y. Catal. Sci. Technol. 2017, 7 (1),
168–180.
(5) Muller, P.; Burt, S. P.; Love, A. M.; McDermott, W. P.; Wolf, P.; Hermans, I. ACS
Catal. 2016, 6 (10), 6823–6832.
(6) Sushkevich, V. L.; Ivanova, I. I.; Taarning, E. Green Chem. 2015.
(7) Shylesh, S.; Gokhale, A. A.; Ho, C. R.; Bell, A. T. Acc. Chem. Res. 2017,
acs.accounts.7b00354.
(8) Pomalaza, G.; Capron, M.; Ordomsky, V.; Dumeignil, F. Catalysts 2016, 6 (12),
203–237.
(9) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A. Catal.
Sci. Technol. 2015.
(10) Kvisle, S.; Aguero, A.; Sneeden, R. P. A. Appl. Catal. 1988, 43 (1), 117–131.
(11) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A.
ChemSusChem 2014, 7 (9), 2505–2515.
(12) Hayashi, Y.; Akiyama, S.; Miyaji, A.; Sekiguchi, Y.; Sakamoto, Y.; Shiga, A.;
Koyama, T.; Motokura, K.; Baba, T. Phys. Chem. Chem. Phys. 2016, 18 (36),
25191–25209.
(13) Di Valentin, C.; Del Vitto, A.; Pacchioni, G.; Abbet, S.; Wörz, A. S. S.; Judai, K.;
Heiz, U. J. Phys. Chem. B 2002, 106 (46), 11961–11969.
(14) Cavalleri, M.; Pelmenschikov, A.; Morosi, G.; Gamba, A.; Coluccia, S.; Martra, G.
In Oxide-based Systems at the Crossroads of ChemistrySecond International
Workshop October 8-11, 2000, Como, Italy; A. Gamba, C. C. and S. C. B. T.-S. in
S. S. and C., Ed.; Elsevier, 2001; Vol. Volume 140, pp 131–139.
(15) Branda, M. M.; Rodríguez, A. H.; Belelli, P. G.; Castellani, N. J. Surf. Sci. 2009,
603 (8), 1093–1098.
(16) Taifan, W. E.; Bučko, T.; Baltrusaitis, J. J. Catal. 2017, 346, 78–91.
(17) Chieregato, A.; Velasquez Ochoa, J.; Bandinelli, C.; Fornasari, G.; Cavani, F.;
Mella, M. ChemSusChem 2014, 8 (2), 377–388.
(18) Fan, D.; Dong, X.; Yu, Y.; Zhang, M. Phys. Chem. Chem. Phys. 2017, 19 (37),
25671–25682.
(19) Sushkevich, V. L.; Palagin, D.; Ivanova, I. I. ACS Catal. 2015, 4833–4836.
(20) Harris, J. W.; Cordon, M. J.; Di Iorio, J. R.; Vega-Vila, J. C.; Ribeiro, F. H.;
Gounder, R. J. Catal. 2016, 335, 141–154.
(21) Boronat, M.; Concepcion, P.; Corma, A.; Navarro, M. T.; Renz, M.; Valencia, S.
Phys. Chem. Chem. Phys. 2009, 11 (16), 2876–2884.
(22) Boronat, M.; Concepción, P.; Corma, A.; Renz, M.; Valencia, S. J. Catal. 2005,
234 (1), 111–118.
(23) Kantorovich, L. N.; Holender, J. M.; Gillan, M. J. Surf. Sci. 1995, 343 (3), 221–
192
239.
(24) Chizallet, C.; Petitjean, H.; Costentin, G.; Lauron-Pernot, H.; Maquet, J.;
Bonhomme, C.; Che, M. J. Catal. 2009, 268 (1), 175–179.
(25) Chizallet, C.; Costentin, G.; Che, M.; Delbecq, F.; Sautet, P.; Surface, D.; Marie,
P. J. Am. Chem. Soc. 2007, No. 19, 6442–6452.
(26) Chizallet, C.; Costentin, G.; Che, M.; Delbecq, F.; Sautet, P. J. Phys. Chem. B
2006, 110 (32), 15878–15886.
(27) Sushko, P. V.; Shluger, A. L.; Catlow, C. R. A. Surf. Sci. 2000, 450 (3), 153–170.
(28) Bailly, M.; Chizallet, C.; Costentin, G.; Krafft, J.; Lauronpernot, H.; Che, M. J.
Catal. 2005, 235 (2), 413–422.
(29) Sushkevich, V. L.; Vimont, A.; Travert, A.; Ivanova, I. I. J. Phys. Chem. C 2015,
119 (31), 17633–17639.
(30) Janssens, W.; Makshina, E. V. V.; Vanelderen, P.; De Clippel, F.; Houthoofd, K.;
Kerkhofs, S.; Martens, J. A. A.; Jacobs, P. A. A.; Sels, B. F. B. F. ChemSusChem
2014, 8 (6), 994–1008.
(31) Larina, O. V.; Kyriienko, P. I.; Trachevskii, V. V.; Vlasenko, N. V.; Soloviev, S.
O. Theor. Exp. Chem. 2016, 51 (6), 387–393.
(32) Zhu, Q.; Wang, B.; Tan, T. ACS Sustain. Chem. Eng. 2017, 5 (1), 722–733.
(33) Golay, S.; Kiwi-Minsker, L.; Doepper, R.; Renken, A. Chem. Eng. Sci. 1999, 54
(15), 3593–3598.
(34) Di Cosimo, J. I.; Dıez, V. K.; Xu, M.; Iglesia, E.; Apesteguıa, C. R. J. Catal. 1998,
178 (2), 499–510.
(35) Phung, T. K.; Busca, G. Catal. Commun. 2015, 68, 110–115.
(36) Taifan, W.; Yan, G. X.; Baltrusaitis, J. Catal. Sci. Technol. 2017, 7 (20), 4648–
4668.
(37) Gao, M.; Jiang, H.; Zhang, M. Appl. Surf. Sci. 2018.
(38) Zhu, Q.; Wang, B.; Tan, T. ACS Sustain. Chem. Eng. 2017, 5 (1), 722–733.
(39) Kotov, S. V.; Nankeva, I. N. Chem. Technol. Fuels Oils 1994, 30 (5), 240–245.
(40) Ho, C. R.; Shylesh, S.; Bell, A. T. ACS Catal. 2016, 6 (2), 939–948.
(41) Mathur, L.; Saddam Hossain, S. K.; Majhi, M. R.; Roy, P. K. Boletín la Soc.
Española Cerámica y Vidr. 2017.
(42) Tavangarian, F.; Emadi, R. Mater. Res. Bull. 2010, 45 (4), 388–391.
(43) Tavangarian, F.; Emadi, R. Ceram. – Silikáty 54 (2), 122–127.
(44) Swanson, H. E.; Tatge, E. J. Res. Natl. Bur. Stand. (1934). 1951, 46 (4), 318–327.
(45) ter Veen, H. R. J.; Kim, T.; Wachs, I. E.; Brongersma, H. H. Catal. Today 2009,
140 (3), 197–201.
(46) Davydov, A. Molecular Spectroscopy of Oxide Catalyst Surfaces; John Wiley &
Sons Ltd: West Sussex PO19 8SQ, England, 2003.
(47) Keturakis, C. Operando Molecular Spectroscopy during Catalytic Biomass
Pyrolysis, Lehigh University, 2015.
(48) Knözinger, E.; Jacob, K.-H.; Singh, S.; Hofmann, P. Surf. Sci. 1993, 290 (3), 388–
402.
(49) Toussaint, W. J.; Dunn, J. T.; Jachson, D. R. Ind. Eng. Chem. 1947, 39 (2), 120–
125.
(50) Palagin, D.; Sushkevich, V. L.; Ivanova, I. I. J. Phys. Chem. C 2016, 120 (41),
23566–23575.
193
(51) Ordomsky, V. V.; Sushkevich, V. L.; Ivanova, I. I. J. Mol. Catal. A Chem. 2010,
333 (1), 85–93.
(52) Bian, S.-W.; Baltrusaitis, J.; Galhotra, P.; Grassian, V. H. J. Mater. Chem. 2010,
20 (39), 8705.
(53) Busca, G.; Lorenzelli, V. Mater. Chem. 1982, 7 (1), 89–126.
(54) Taifan, W.; Boily, J.-F.; Baltrusaitis, J. Surf. Sci. Rep. 2016, 71 (4), 595–671.
(55) Lercher, J. A.; Grijndling, C.; Eder-Mirth, G. Catal. Today 1996, 27, 353–376.
(56) Lavalley, J. C. Catal. Today 1996, 27, 377–401.
(57) Vimont, A.; Thibault-Starzyk, F.; Daturi, M. Chem. Soc. Rev. 2010, 39 (12), 4928–
4950.
(58) Busch, O. M.; Brijoux, W.; Thomson, S.; Schüth, F. J. Catal. 2004, 222 (1), 174–
179.
(59) DeWilde, J. F.; Chiang, H.; Hickman, D. A.; Ho, C. R.; Bhan, A. ACS Catal. 2013,
3 (4), 798–807.
(60) Da Ros, S.; Jones, M. D.; Mattia, D.; Pinto, J. C.; Schwaab, M.; Noronha, F. B.;
Kondrat, S. A.; Clarke, T. C.; Taylor, S. H. ChemCatChem 2016, 8, 1–12.
(61) Joshi, S.; Kalyanasundaram, S.; Balasubramanian, V. Appl. Spectrosc. 2013, 67
(8), 841–845.
(62) Kim, J.-W.; Lee, H.-G. Metall. Mater. Trans. B 2001, 32 (1), 17–24.
(63) Echterhoff, R.; Knözinger, E. J. Mol. Struct. 1988, 174, 343–350.
(64) Echterhoff, R.; Knözinger, E. Surf. Sci. 1990, 230 (1), 237–244.
(65) Angelici, C.; Meirer, F.; van der Eerden, A. M. J.; Schaink, H. L.; Goryachev, A.;
Hofmann, J. P.; Hensen, E. J. M.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS
Catal. 2015, 5 (10), 6005–6015.
194
Chapter 6
Role of transition metal promoters (Cu, Zn)
on MgO/SiO2 catalyst for Lebedev reaction
Abstract ..........................................................................................................................194
1. Introduction .......................................................................................................195
2. Computational results ...................................................................................... 199 2.1. Model catalyst selection and analysis ...................................................... 199
2.2. Reactive intermediates ............................................................................. 206
2.3. Potential energy surfaces .......................................................................... 210
3. Experimental results .........................................................................................214 3.1. Catalyst characterization .......................................................................... 214
3.2. Steady state catalytic performance and acid/base chemistry of the
catalyst active sites .............................................................................................. 222
3.3. Active sites under operating conditions ................................................... 226
3.3.1. Temperature programmed infrared spectroscopy measurements (TP-
DRIFTS) ............................................................................................................. 226
3.3.2. In-situ UV-Vis DRS study of MgSi catalysts .......................................... 230
3.3.3. Operando XAS studies of Cu, Zn-promoted MgSi catalysts ................... 232
3.3.3.1. Operando XANES and EXAFS of Cu-promoted MgSi catalyst ......... 232
3.3.3.2. Operando XANES and EXAFS of Zn-promoted MgSi catalyst ......... 241
4. Conclusion ......................................................................................................... 245
Supporting Information ............................................................................................... 247
References ...................................................................................................................... 266
Abstract
Electronic structure and reactivity of Cu- and Zn-promoted wet kneaded MgO/SiO2
catalysts was interrogated during ethanol reaction to 1,3-BD. A multimodal nature of
characterization, including in situ or operando X-ray, electron, light spectroscopies and
steady state reactivity measurements demonstrated critical new information on the
195
temporal evolution of the catalyst active sites including key measurements performed
operando using synchrotron source (EXAFS and XANES). In situ DRIFT spectroscopy
allowed to decouple the aldol condensation and dehydrogenation reactive steps due to the
promotion with enhanced ability to carry out aldol condensation, as correlated with the
steady state reactivity experiments. . In situ UV-Vis spectroscopy presented a complex
picture of the adsorbates with π- π* electronic transitions due to the allylic cations, cyclic
or aromatic species while also suggesting oligomeric CuO species were formed. Operando
X-ray measurements combined with ab initio multiple scattering modelling performed as
a function of temperature identified a new transient intermediate assigned to a 4-fold
coordinate Cu species that was key leading to increase in Cu-Cu bond number. For the
first time, two types of Zn bonds, namely Zn-O and Zn-Mg, were identified during X-ray
analysis under operating conditions. With Zn nearly 6-coordinated when in the vicinity of
Mg while Zn-O species coordinated to nearly 4 nearest neighbors. The data suggest that
such supported catalyst deactivation might proceed not only via carbon coking mechanism
but also through the dispersed Cu site diffusion and growth due to the nearest neighbor
oxygen atoms loss. The results presented suggest intermediates for
segregation/deactivation mechanisms for a broader set of supported Cu and Zn catalysts
used for alcohol upgrading catalytic reactions.
1. Introduction
Catalytic conversion of ethanol to 1,3-butadiene (1,3-BD) is a promising green and
renewable route for obtaining a commodity chemical that does not utilize a conventional
196
petroleum-based feedstock.1 The feedstock and technological process landscape in 1,3-BD
production is undergoing changes due to the distinct industry shift from oil to C4
hydrocarbon lean shale gas.2 To this regard, ethanol is a very interesting platform molecule
due to its steadily increasing production from biomass.1 Two classes of catalysts have been
used for ethanol conversion to 1,3-BD, namely ZrO2-based and MgO/SiO2-based (Lebedev
catalyst).3 The former have thoroughly been investigated using a combination of
computational and spectroscopic methods4,5 while the latter lack suitable spectroscopic
characterization.3 The overall reaction mechanism on MgO/SiO2 is currently debated3,6–8
and several recent attempts have been made to elucidate it.6,9–12 These studies pointed
towards aldol condensation as the most energetically favorable C-C bond formation
mechanism, except for Chieregato et al. who suggested that C-C bond was formed via
interaction of ethanol/acetaldehyde through a stable carbanion intermediate.9 The latter
mechanism was suggested to take place on pure, basic MgO sites based on a combination
of infrared spectroscopy and theoretical DFT results.9 The rate-determining step was found
to be ethanol dehydrogenation6,11 since an efficient dehydrogenating site was not present
in MgO/SiO2 catalysts. This suggests that an effective catalyst must possess
multifunctional, i.e. acidic, basic and redox sites. MgO/SiO2 catalysts are promoted with
transition metal (oxides) to improve their dehydrogenation capability2,13–17 where the
choice of transition metal used as a promoter is determined by its dehydrogenation
capability.18–20 Au,21,22 Ag,23,24 and Cu25,26 have been utilized to enhance the 1,3-BD
yield.2,27,28 Zn is another promoter that has been utilized to improve the yield of 1,3-
BD.13,15,29–31 The promotional effect was reported to originate from the improved
availability of both Lewis acid sites and redox sites.3,15 While Au and Ag promoters present
197
economic constraints due to their high costs, Cu and Zn are relatively inexpensive and
present an alternative for an efficient catalyst design. The work reported here provides
new insights on the structure and reactivity of these sites under operating conditions.
Several theoretical and ultra-high vacuum (UHV) studies have been conducted on
Cu-based catalysts to determine the structure of the active sites32–39 but very few under
operating conditions. UHV characterization and DFT revealed formation of isolated or
clustered Cu0 phases on the MgO surface32,33 or a solid solution that contains Cu-Mg and
Cu-O-Mg bonds.34 For instance, on a perfect MgO (100) surface, DFT calculations showed
that a single Cu adatom prefers to bond with a surface O atom with the possibility to
spillover Cu.32,33 Various cluster sizes of Cu (dimers, trimers, and tetramers) were observed
depending on the surface coverage.32 The formation of reduced Cu clusters on the surface
was confirmed by Colonna et al. where Cu clusters, as evident by Cu-Cu bond length (2.55
Å), were observed as a thin layer on MgO using XANES during the UHV evaporation-
deposition synthesis.35 In a separate study, in addition to the observed Cu atoms on the
MgO surface, both UHV XANES and DFT identified the formation of a solid solution
between Cu and MgO that decreased the reactivity of the catalyst toward H2S and SO2
decomposition when compared to the supported Cu atom.36,37 Larger charge transfer
resulting in a strong ionic bond was observed when Cu was coordinated next to a defective
MgO surface.38,39 This shorter bond was due to the electron stabilization provided by the
Cu atom.38,39 UHV XANES of several transition metal-promoted MgO catalysts utilized
for CH3OH and RCH2Z (where R=H and CH3, Z=CN, COR’, and COOR”) coupling
reactions confirmed the formation of Cu-MgO solid solution at 80 K and suggested that an
octahedral coordination of the Cu species due to the pre-edge peak associated with 1s3d
198
transition was very small. This observation was accompanied by the extended X-ray
absorption fine structure (EXAFS) analysis of the Cu-O and Cu-Mg atomic distances, 2.01
Å and 2.98 Å respectively, suggesting the formation of solid solution between Cu and MgO.
Interestingly, all promoted MgO catalysts that showed worse catalytic activity toward the
coupling reaction of CH3OH and RCH2Z (R=H and CH3, Z=CN, COR’, and COOR”) were
those that formed a solid solution with MgO.34 Applicable to this study is in-situ (operando)
characterization of a Cu-promoted catalyst for relevant alcohol reactions, such as methanol
formation from syngas,40 ethyl acetate production from ethanol41 and ethanol steam
reforming.42 Cu-containing ternary oxide catalysts, e.g. Cu/ZrO-SiO2, CuMgAlOx and
Cu/MgO-SiO2, were utilized for these and were well characterized.28,43,44 Operando and
in-situ characterization of these supported catalysts showed that Cu species could be
present as both Cu2+ ions and CuO - the latter exhibiting lower-strength interaction with
the SiO2 support,25,26 as a solid solution in the case of Cu-MgO/SiO228 and
Cu/ZnO/Al2O3,45 as dimeric structures in the case of CuMgAl hydrotalcite catalysts44 or as
reduced species as in the case for CuCrOx and CuZrSiOx catalysts.43,46 This suggests variety
of active copper sites can be present under operating conditions28,43–45 but very few studies,
notably Angelici et al.,26,28 attempted to decouple their reactivity during 1,3-BD formation
or investigate the temperature effect on Cu site composition under reactive conditions.28
ZnO/SiO2 has been used as a model catalyst for many reactions, such as water-gas shift and
methanol formation reaction,47 but X-ray based catalytic site characterization during
ethanol-to-1,3-BD are not existent to the best of our knowledge.13,15,16 In-situ XAS and
UV-Vis of this catalyst further showed the relevance of the precursor drying steps during
the synthesis and that Zn was present both as a silicate (hemimorpite) and ZnO bulk phase
199
at 10% Zn loading.47 Ambient UV-Vis and TEM studies of a 1% ZnO/MgO catalyst
demonstrated the formation of a highly-dispersed ZnO layer which had high activity for
CO oxidation, affected by the quantum-confinement effect.48
In this work, we performed a comprehensive characterization on both Cu- and Zn-
promoted MgO/SiO2 catalysts. Details on the acid-base sites upon promotion with Cu and
Zn, implications for the reaction mechanisms, as well as thorough infrared, UV-vis,
electron and X-ray-based analysis of Cu and Zn local structure before, after, and during the
reaction was elucidated. Complementary, if not contradictory, conclusions were reached
for Cu-promoted MgO/SiO2 to those available in the literature28 while completely new X-
ray data insights were obtained for Zn-promoted MgO/SiO2 catalysts under operating
conditions.
2. Computational results
2. 1. Model catalyst selection and analysis
Plenty literature of Cu-doped MgO catalysts characterization is available, both on
computational and experimental studies. DFT calculation of small Cu cluster supported on
a perfect MgO (100) surface revealed that for a single Cu adatom, the preferential
adsorption site was on top of an O atom, whereas adsorption on a hollow site represented
a saddle point for Cu spillover.32,33 For a dimer Cu, there were two minimum states
available with close optimized energies, parallel and linearly perpendicular to the surface.
Two states were observed for the first case; one configuration where the dimer bond is 2.25
Å (stretched from 2.25 Å in its free form), and another one where the bond length is
stretched even further, 2.34 Å. The linearly perpendicular states, however, had the single
200
adatoms’s Cu-O bond length, and free dimer’s Cu-Cu bond length. Trimer Cu and tetramer
Cu clusters preferred linear and rhombus geometry, respectively.32 Cu/MgO DFT model
had been used by Jose Rodriguez and coworkers.36,37 When Cu was embedded into the
surface, substituting an Mg atom with lower coordination, the catalyst was less reactive
compared to when the Cu atom was adsorbed freely on the surface. When a SH or S
molecule was adsorbed on the embedded Cu atom, the interaction was so strong that it
pulled the Cu atom out of the surface plane.36,37
A significantly larger charge transfer was observed by Zhukovskii et al. and
Matveev et al. when the metal was adsorbed on a defective MgO surface, meaning that the
bonding was more ionic than that on the perfect surface.38,39 The distance between Cu atom
and the surface had decreased as well, from ~2 Å to 1.62 Å. This stronger bonding
originated to the lower coordination atom, which would behave more like ions due to the
lack of electronic relaxation. The defect sites used here are both Fs and Fs+ sites, where an
oxygen atom was removed from a perfect surface, along with a number of electrons
accordingly to create the oxygen vacancy.
Experimental data carried out by Asakura and Iwasawa provided a different
insight.34 On a doped MgO catalyst, prepared using wet impregnation method, XANES
spectra suggested an octahedral coordination for Cu+ ions, deduced from the fact that the
pre-edge peak of the spectra which was assigned to the 1s-3d transition was very small.
The EXAFS spectra for Cu+ ion, revealed that the M-O and M-Mg distances were observed
to be 0.201 nm and 0.298 nm, respectively, away from the lattice constant of MgO, which
further suggested that the Cu+ ion would substitute an Mg site, i.e. supplanted into the
lattice.34 Colonna et al.35, however, failed to replicate the XAS experiments when the
201
surface was prepared using ion evaporation-deposition method in an UHV chamber. On
monolayer coverage, the EXAFS data evidenced a Cu-Cu distance close to that of the bulk
metal, due to the weak film-substrate interaction. Copper was also observed to be in its
reduced state, and further, XANES spectrum showed a coordination number similar to the
bulk value, indicating that the Cu+ ions grew as a cluster on the MgO substrate.35 Different
preparation led to different geometry, as observed by Pascual et al., which carried out X-
ray measurement on Cu-doped MgO using arc fusion method, where MgO was melted
before the dopant is mixed as CuO.49 Both EXAFS and Ab initio calculation showed that
the crystal is in D4h geometry, associated with a compression of the original octahedron
along the z-axis, indicating that the ions substituted Mg sites in the lattice.
Although the literature on Cu/MgO catalysts is very well-established, very limited
study is available on promoted MgO/SiO2 catalyst. Complications on how Cu would be
added to the catalyst were brought upon by the presence of a second support material, i.e.
SiO228 or Al2O3
50. The Cu could be present either as a surface species on SiO2, as in the
case of Cu/SiO2 catalysts28, or as a substitutional dopant, replacing Mg as in the case of the
Cu/MgO catalysts. MgO/SiO2, the Lebedev catalyst, was studied by Angelici et al.28 Ex-
situ XANES, EXAFS, FTIR, XRD, TEM, XPS, and UV-Vis showed that Cu species were
not in planar geometry and were suggested to be located at crystal lattice sites. Formation
of small (CuO)x clusters on MgO-containing materials was also advocated experimentally.
XANES and EXAFS had also demonstrated the octahedral coordination of CuO and Cu-
Mg bond distance of 0.298-0.302 nm, similar to the Mg-Mg distance in periclase phase.28
Hydrotalcites (MgxAlyOz) were another class of catalysts that were routinely studied for
ethanol upgrading. High resolution NMR study of Cu-promoted MgxAlyOz was extensively
202
studied and revealed that on low loading, Cu preferentially substituted for lattice Mg in the
hydrotalcite structure, while at higher Cu content, the transition metal was also present on
the surface as a bulk oxide.50
Studiy on the routinely used ZnO/SiO2 catalysts showed that zinc was mostly
present as Zn silicates in addition to small amount of ZnO bulk phase on the surface.47
Extensive characterization of the catalyst was carried out with XAS, DRIFTS, and UV-Vis.
Depending on the drying temperature, the small amount of bulk ZnO phase was formed on
top of the catalyst after calcination. The calcination mostly resulted in the formation of Zn-
silicate, hemimorphite in particular.47 However, when MgO was added to the catalyst, i.e.
Zn-doped Lebedev catalyst, more possibilities are now available, including the
substitutional doping of Mg by Zn, formation of ZnO bulk phase on either MgO or SiO2,
surface species formation on either support material, or preferential formation of Zn-
silicate. Colloidal suspension synthesis method of 1% ZnO/MgO was shown to result in a
very highly dispersed ZnO layer on top of MgO support, as confirmed by UV-Vis and
TEM.48 However, the catalyst synthesized in this study did not show the characteristic ZnO
band gap and operando XANES-EXAFS characterization of the catalyst suggested that the
local structure environment is very similar to Cu, indicating interaction with MgO, instead
of SiO2, resulting in a solid solution (vide infra). Hence, the model selection for both
catalysts was chosen to be CuMgO and ZnMgO, with both transition metals to
substitutionally dope an Mg atom. These models serve as a first approximation to the
catalysts’ model, simplifying the SiO2 support effects, further eliminating the contribution
of Mg-O-Si linkages and the accompanying hydroxyl groups.
203
Table 6.1 Different configurations tested for Zn(Cu)/MgO model catalysts. Various dopant
location was chosen between the top and second layer, and compared for energy and Bader
charge.
Scheme Configuration Dopant location
Cu-doped Zn-doped
Energy
(eV)
Bader
Charge
Energy
(eV)
Bader
Charge
Cu(Zn)-
1 Cu3C-top layer
-659.48
(0.00) 0.72
-658.64
(0.00) 1.05
Cu(Zn)-
2 Cu5C-top layer
-659.28
(0.20) 0.82
-658.42
(0.22) 1.05
Cu(Zn)-
3 Cu4C-top layer
-659.22
(0.27) 0.82
-658.44
(0.20) 1.09
Cu(Zn)-
4
Cu5C-second
layer
-658.97
(0.51) 0.94
-658.15
(0.49) 1.16
Cu(Zn)-
5
Cu5C-second
layer
-658.89
(0.60) 0.92
-658.10
(0.54) 1.08
Cu(Zn)-
6 Cu4C-top layer
-659.13
(0.36) 0.81
-658.37
(0.27) 1.05
Cu(Zn)-
7 Cu4C-top layer
-659.45
(0.04) 0.91
-658.52
(0.12) 1.08
Cu(Zn)-
8
Cu5C-second
layer
-658.83
(0.65) 0.92
-658.09
(0.55) 1.08
204
Cu(Zn)-
9
Cu5C-second
layer
-658.73
(0.75) 0.96
-658.03
(0.61) 1.13
Cu(Zn)-
10
Cu5C-second
layer
-658.75
(0.73) 0.93
-658.08
(0.57) 1.15
Table 6.1 shows the permutations tried for both dopants, i.e. Cu and Zn. The
original kink model from Chapter 3 was used and modified with dopants substitutionally
dope the catalyst’s surface at different Mg atom locations. From all of the tried models, Cu
(Zn)-1 possesses the lowest electronic energy. The Bader charge for the transition metal
atom for each configuration was also calculated, with charge values of +0.72 and +1.05 for
Cu and Zn. To discuss the effects of both Cu and Zn on the MgO model catalysts, Bader
charges of the neighboring atoms were calculated as well and compared to the undoped
MgO catalyst, shown in Figure 6.1.
Figure 6.1. Local structure analysis of (a)MgO, (b)Cu-MgO, and (c)Zn-MgO. The
Bader atomic charge on each atom is indicated by the boldfaced numbers.
(a) (b) (c)
205
The oxygen atoms neighboring both transition metals exhibited less negative
charges than that of unpromoted MgO, with total charges of the 3 O atoms amounting to -
4.91, -4.31, and -4.4 e for MgO, Cu-MgO, and Zn-MgO, respectively. The decreased
atomic charges of O atoms neighboring the corner atoms indicates that the introduction of
both transition metal atoms had lowered the basicity of the O atoms. The lowered basicity
of the oxygen atoms neighboring the transition metal atoms was also observed both
computationally and experimentally using XPS on Zn-promoted talc by Baba’s group.13
The neighboring Mg atoms and other oxygen atoms that are distanced from the transition
metal dopants did not show any difference from the pristine MgO, which suggests that this
lowered basicity effect is very localized, and hence calculation should be focused on this
region.
The atomic charges observed for both Cu and Zn atoms are similar to what was
previously observed on CuO51 and ZnO52 surfaces. Bader charges for CuO and ZnO are
typically +0.57 – +0.84 and +1.13 – +1.20, respectively.51,52 The very positive atomic
charges for both means that these transition metal atoms are almost fully ionized in these
model catalysts. The lowered atomic charges of both Cu and Zn from the ionized Cu+1 and
Zn+2 are due to the charge transfer from the neighboring O atom. For Zn, the charge transfer
is very straightforward, since both Mg and Zn possess +2e charge. The charge transfer for
the corner Mg (Zn) from clean MgO (Zn-MgO) surface is equal to the difference between
the Bader charge value and the charge of isolated Mg (Zn). These values are 0.4e and 0.95e
for Mg and Zn, respectively. The difference of this value is equal to the difference of the
summed charges on the neighboring O atoms between MgO and Zn-MgO, i.e. -4.91 and -
4.4. The origin of charge density change is also acknowledged by Baba’s group for the case
206
of Zn/talc.13 For Cu, however, this analysis fell apart, since the atomic charge of Cu is +1e,
while Mg is +2e, and the lowered total charges on neighboring O atoms is different from
the difference between Mg and Cu atomic charges.
2. 2. Reactive intermediates
Zn-1A Zn-1B (TS) Zn-2A Zn-2B (TS)
Zn-1C Zn-2C
Zn-3A Zn-3B (TS) Zn-4A Zn-4B (TS)
Zn-3C Zn-4C
207
Cu-1A Cu-1B (TS) Cu-2A Cu-2B (TS)
Cu-1C Cu-2C Cu-2D
Cu-3A Cu-3B (TS) Cu-4A Cu-4B (TS)
Cu-3C Cu-4C
All optimized structures, including maxima and minima, are shown in Figure 6.2,
while the corresponding electronic energies and Gibbs corrected free energies were
tabulated in Table 6.2. The structures were optimized on the model Zn and Cu-doped MgO
Figure 6.2. All stable intermediates and transition states calculated following the reaction
pathways. (1A-1C): ethanol dehydrogenation to acetaldehyde; (2A-2C): ethanol
dehydration to ethylene; (3A-3C): C-C bond formation step in acetaldehyde aldol
condensation to 3-hydroxybutanal (acetaldol); (4A-4C): C-C bond formation step in Prins
condensation of acetaldehyde and ethylene. Calculations are carried over Zn/MgO model
catalysts (prefix: Zn), and Cu/MgO model catalysts (prefix: Cu).
208
catalysts. To produce a comparable result, the optimization was done on the same site, i.e.
corner Zn (Cu) atoms and the neighboring O atoms. The reaction steps that were studied
are the key steps in both Prins and aldol mechanisms, i.e. dehydrogenation, dehydration,
and C-C bond formation of both aldol and Prins condensation pathways.
Table 6.2. Referenced electronic and corrected Gibbs free energy for each species over
MgO, Cu/MgO, and Zn/MgO catalysts.
For the dehydrogenation step, the intermediates are shown by Cu(Zn)-1A to 1C.
On all MgO, Cu-MgO, and Zn-MgO, ethanol undergoes spontaneous dissociation to give
surface ethoxide and a hydroxyl group. There was no visible difference from the model,
and the electronic energies are not significantly different, i.e. ~35-40 kcal/mol. The Gibbs
corrected free energies, however, are very different with Cu-1a is now the most stable
species and Zn-1a being the least. The transition state for the dehydrogenation reaction did
look very similar to the optimized structure on bare MgO catalyst (Chapter 3); the H-H and
H-lattice O4C distances are all of similar values.6 The energetic values, however, differ a
Species MgO Cu/MgO Zn/MgO
E
(kcal/mol)
G
(kcal/mol)
E
(kcal/mol)
G
(kcal/mol)
E
(kcal/mol)
G
(kcal/mol)
1a -40.6 -13.50 -37.33 -18.72 -35.75 -8.10
1b 3.9 26.20 14.33 32.12 12.36 36.23
1c -16.6 5.60 6.03 14.22 4.93 18.59
2a -39.9 -10.50 -25.66 -1.96 -31.45 -4.30
2b -2.6 23.00 4.97 21.48 5.99 31.54
2c -32 -15.30 -26.19 -1.65 -26.45 -8.22
3a -22.5 -23.70 -5.40 -14.32 -4.41 -4.21
3b -15.9 -7.60 3.07 3.22 -4.24 0.75
3c -19.4 -10.80 -2.89 -2.45 -8.72 -3.51
4a 11.6 -0.20 21.13 7.78 21.65 13.80
4b 25.6 28.60 34.17 31.13 33.81 35.49
4c -34.8 -22.90 -0.31 9.29 -27.25 -15.61
209
lot from the bare MgO, ~6-10 kcal/mol higher than the transition state found for the bare
MgO. The very high values for both electronic and Gibbs’ corrected energies indicate that
these structures are very unstable. This observation is followed as well with the final state
of the reaction, i.e. the adsorbed acetaldehyde and hydrogen as a product of the reaction
when hydrogen is fully formed.
Ethylene production is an unwanted side reaction that accompanies 1,3-BD reaction
pathway, and was even linked to as a reactive intermediate in the Prins mechanism.53 Over
MgO catalyst, ethanol to dehydration possessed lower activation energy than
dehydrogenation, which confirms the experimental evidence.6 Over Zn(Cu)-MgO, the
initial state of this reaction, i.e. ethanol chemisorption, is very similar to the state observed
on MgO catalyst. Common structural parameters, such as the supposedly elongated C-O
bond, Cu(Zn) distance from the ethoxide O atom, and the bonding between O and planar
Mg5C are very similar to those in the bare MgO catalyst. Energetically, these intermediates
are much less stable than the MgO-bound species, differing ~8-10 kcal/mol. These
instabilities were also observed in the transition state and the final state of the reaction. Cu-
MgO, in particular, exhibits a very high degree of affinity toward the ethylene product, as
shown by the stabilized carbanion in Cu-2C. This intermediate then undergoes a C-Cu bond
breaking transition step (not optimized here) to give off ethylene as the final product, i.e.
Cu-2D. Cu-2D possessed referenced electronic energy of -27.54 kcal/mol, only 1 kcal/mol
more stable than Cu-2C.
Probably the most debatable step in the reaction mechanism is the C-C bond
formation. Aldol condensation had been widely accepted as the most possible mechanism,
and we optimized all the transition states over the doped model catalysts, following the
210
calculated structure over bare MgO.6 Over the doped catalysts, the intermediates, i.e.
minima, are all less stable than that on MgO. The stability achieved by the system when
two acetaldehyde molecules are coadsorbed, i.e. initial state, on the surface is not replicated
well on the doped MgO. Surprisingly, the activation energy of this step is much lower for
Zn-MgO than Cu-MgO and bare MgO (discussed in detailed manner in Section 2.3). The
viability of Prins mechanism again is questioned computationally. On all the optimized
structures, similar activation energies were observed, i.e. ~21-29 kcal/mol. The
computational method also favors the reaction step thermodynamically, with all catalysts
giving exergonic reaction for the Prins step. This Prins mechanism, however, should be
treated carefully, since experimental evidence pointed out that ethylene formation is very
exclusive from 1,3-BD formation (Chapter 5).
2. 3. Potential energy surfaces
The potential energy surfaces for all computed reaction steps are presented in
Figure 6.3 and 6.4, while the activation barriers and Gibbs’ free energies of reactions are
tabulated in Table 6.3. Based on our calculation, promotion with transition metals, i.e. Cu
and Zn, did not achieve the intended lowered activation energy of ethanol dehydrogenation.
Rather, these activation barrier increased for the case of Zn and Cu to 44.33 and 50.84
kcal/mol, respectively. For dehydration, dehydration activation energies are consistent with
bare MgO, with similar values attained, especially for Zn-MgO. With the presented
activation energies for both C2 reactions, i.e. dehydrogenation and dehydration, the
reaction mixture at low temperature, will mostly consist of ethylene, instead of
acetaldehyde, for all catalysts.
211
Table 6.3 Activation energy and thermodynamics consideration for key steps during
ethanol conversion to 1,3-butadiene over MgO, Zn/MgO, and Cu/MgO catalysts.
Reaction steps ΔGA (kcal/mol) ΔGRx (kcal/mol)
MgO Zn/MgO Cu/MgO MgO Zn/MgO Cu/MgO
Dehydrogenation 39.60 44.33 50.84 24.61 26.68 32.94
Dehydration 33.47 35.84 23.44 -4.87 -3.92 0.31
Aldol C-C 16.10 4.96 17.55 19.07 0.70 11.87
Prins C-C 28.75 21.69 26.39 -22.74 -29.42 -1.87
Very surprisingly, aldol condensation is much more favorable over Zn-MgO, with
the very low activation energy, as well as the lowered Gibbs’ free energy of reaction value
of 0.70 kcal/mol, which is almost aergonic. This lowered energetic barrier might be due to
the duality of Zn, which presents as both redox site and as a Lewis acid site, which
supposedly boost both dehydrogenation and aldol condensation step during the reaction.3
Another unexpected observation is that of Prins mechanism, which continues to show
viability computationally. Although this step does not have experimental ground, the
theoretical calculation shows that this step is very feasible. According to our calculation,
this step and ethanol dehydration are the two steps that have net positive rate constant
(Chapter 3).6 However, the calculation was carried out over model MgO catalyst, without
considering the presence of SiO2 and OH groups, therefore eliminating the other possible
sites, such as the open and closed Lewis acid sites of Mg-O(H)-Mg and Mg-O(H)-Si, which
were shown to actively catalyze the whole reaction steps (Chapter 5).54
The non-existence of hydroxyl groups in this idealistic model leads to very strong
Lewis acid-base pairs, which is accentuated by the electron-deficient sites such as kink,
corners, and edges. These highly unstable sites are typically stabilized by hydroxyl groups,
and leads to softer acid-base pairs, due to the more distributed electron density.
212
Experimentally, the ethanol dehydrogenation step requires a weak basic site on the catalyst,
as shown by the CO2 and propionic acid cofeeding experiments, in-situ titration with CO2
and pyridine, which is also supported by previous investigation.2 The strong acid-base pairs
used throughout the calculation inevitably stabilize electron rich or deficient structures,
exhibited by the stabilized ethylene-acetaldehyde transition state that leads to the formation
of a C4 oxygenate.6 Furthermore, the proton abstraction steps that follow resulted in a
carbanion, which is stabilized by the presence of corner Mg3C2+, which is undoubtedly a
very strong electron acceptor.6
Another concerning discrepancy with experimental results are the dehydrogenation
step. The reaction mechanism is widely believed to be dictated by ethanol dehydrogenation.
The subsequent aldol condensation, dehydration, and MPV reduction steps were shown to
be facile and spontaneous on unpromoted MgO/SiO2 catalysts.11 Promotion with Cu14 and
Zn,15 among other transition metals,2,16,27 are intended to lower the ethanol
dehydrogenation step and to shift the rate-limiting step to MPV reduction step, which is
already considerably fast. The calculated activation energies for this step on the Cu-MgO
Figure 6.3. Potential energy surface for ethanol (a)dehydrogenation and
(b)dehydration over MgO, Zn/MgO, and Cu/MgO catalysts. (●)MgO, (■) Cu-MgO,
(♦) Zn-MgO.
213
and Zn-MgO are, however, very similar to the bare MgO catalyst. On Cu-MgO, the
activation energy is even ~10 kcal/mol more than the bare MgO. There are two possible
sources of disagreement that is inherent to the model selection and the calculation nature.
DFT calculations on transition metals, especially first row transition metals such as Cu and
Zn, have to be treated carefully. Different DFT functionals had led to large variations in
energies, of 20 kcal/mol or more.55 The pure DFT functional used in this calculation (PBE)
is known to overestimate the stability of low-spin forms. Improvement is usually achieved
by including HF exact exchange, in expense of the prohibitively expensive calculation time,
especially in the large system used for this study.55 The assumption that the whole reaction
steps are carried out on one site is oversimplification of this complex system. While this
might be true on bare MgO, dehydrogenation over transition metal-promoted typically
occurred on an isolated transition metal sites,2,13,28 and the subsequent steps are over the
Mg-O(H)-Si or Mg-O(H)-Mg, which will be shown in the next sections.
Figure 6.4. Potential energy surface for first C-C bond formation via (a) acetaldehyde
aldol condensation and (b) Prins reaction between acetaldehyde and ethylene over
MgO, Zn/MgO, and Cu/MgO catalysts.
214
3. Experimental results
3. 1 Catalyst characterization
The transition metal content in each catalyst was determined using both ICP-OES
and XPS to infer bulk and surface concentration, respectively. Interesting agreement was
found between the two characterization methods with ICP-OES determined Cu and Zn
content of 0.8 % and 2.5 % virtually agreeing with those determined by XPS of 0.9 % and
2.7 % for each catalyst. These Zn and Cu concentrations are close to the intended high
selectivity loading.14,15 The starting support material, i.e. wet-kneaded MgO/SiO2,
possessed surface area of 120 m2/g which was much lower than the fumed SiO2 used (332
m2/g). This lowering of the surface area has previously been observed by several other
groups14,27 and explained by the dispersion of low surface area MgO over SiO2. Promoting
the MgO/SiO2 samples with transition metals led to the increase in the surface area. Zn and
Cu-promoted samples exhibited surface area of 135 and 191 m2/g, respectively. This
significant enhancement of the catalyst surface area was not observed by Janssens et al.27
Rather, Ag-promoted samples were shown to considerably lower the surface area of their
calcined mesoporous support while in this study we used uncalcined hydroxide precursor.
This increase in surface area was likely due to the impregnation step which was done before
the support was calcined. The effect of calcination-impregnation order has previously been
observed by Da Ros et al. with ZrZn-promoted MgO/SiO2 catalysts.16 This suggests that
the metal promoters deposited via impregnation might act also as textural promoters, in
addition to being electronic promoters.
215
X-ray diffraction (XRD) pattern of the two promoted catalysts – CuMgSi and
ZnMgSi – acquired under ambient conditions are shown in Figure 6.5 together with the
unpromoted MgSi. The unpromoted sample exhibited prominent peaks at 37.4, 43.5, 63,
75 and 79° which were due to the periclase MgO. Amorphous silica was also present in the
XRD pattern as evidenced by the broad band in the lower 2θ of 20-30° region. The wet-
kneading between MgO and SiO2 did not produce new bulk crystalline phases in agreement
with Angelici et al.54 Magnesium silicate hydrate phase was previously observed by
Shylesh et al. when MgO/SiO2 catalyst was synthesized by impregnating Mg precursor on
silica.2 Careful examination on the XRD pattern showed that Zn significantly enhanced the
intensity of the MgO peaks suggesting changes in its crystalline structure. For reference,
several concentrations of ZnSi and ZnMg were prepared and also analyzed with XRD
(Figure S6.1). ZnSi showed no new crystalline phases being formed up to 5% loading
while ZnMg also revealed no new crystalline phases were formed at loading up to 10%.
The Cu-promoted catalyst showed no change when compared to the support itself other
than the peak broadening of the MgO periclase structure. However, no new peaks appeared
Figure 6.5. Comparison of XRD patterns between CuMgSi, ZnMgSi, and MgSi.
216
in the Cu-promoted sample as they would appear even at low loading on individual SiO2-
support (Figure S6.2).28 This also showed that Cu promoter was well dispersed on the
catalyst surface with no detectable oxide nanoparticle formed on the surface.28
Figure 6.6 shows DRIFT spectra for dehydrated metal-promoted catalysts in the
OH region, as well as that for the binary catalyst component compounds (ZnSi, ZnMg,
CuSi, CuMg). The promoted MgSi catalysts show similar spectral features to the
unpromoted MgSi. Detailed assignments of the four native OH groups can be found in the
previous work.11 Briefly, there are four prominent peaks on an MgO/SiO2 catalyst, i.e. 3745
cm-1 assigned to both isolated MgO and silanol groups, 3725 and 3705 cm-1 ascribed to
Mg-OH-Si with different OH coordination numbers and 3680 cm-1 peak assigned to a
magnesium silicate species. Promoting the MgSi with Cu or Zn significantly reduced and
broadened the native silica and the WK-signature peaks, i.e. isolated silanol at 3745 cm-1
and Mg-O(H)-Si group at 3680 cm-1. This suggests that both transition metal promoters,
Cu and Zn, interact strongly with this OH group as well. Displacement with Zn further
Figure 6.6. In-situ dehydrated DRIFTS of OH region of MgSi, CuMgSi, and ZnMgSi.
Spectra were taken at 100°C under He flow after pretreatment at 500°C for 1 hour.
Spectra were offset for clarity.
217
results in a new OH site, as shown by the emergence of a peak at 3760 cm-1, which was
previously assigned to the isolated hydroxyl group of MgO.11,56 This highly isolated
hydroxyl group might form from broken Mg-O-Si linkages due to the introduction of Zn
suggesting Zn interaction with O-Mg.
The coordination and oxidation states of the metal promoters are further
characterized by in-situ UV-Vis DRS under dehydrated conditions. Figure 6.7a shows a
comparison between the Cu-promoted (CuMgSi) catalyst, MgSi and reference binary
materials, such as CuMg, CuSi and bulk CuO. UV-Vis DRS spectra of the bulk CuO is
characterized by the presence of a charge transfer (CT) peak at ~251 nm and a peak at 570
nm. The CT peak is assigned to the ligand-to-metal CT (LMCT) from O2- to Cu2+ in
octahedral coordination.44 The peak at 570 nm can be assigned to either surface plasmon
resonance from Cu0 or contributions from d-d transition.57 Furthermore, a peak at 235 nm
Figure 6.7 In-situ UV-Vis DRS spectra of (a) dehydrated CuMgSi catalyst referenced
with Cu/MgO (CuMg), Cu/SiO2 (CuSi), CuO, and MgSi; (b) dehydrated ZnMgSi
catalyst referenced with Zn/MgO (ZnMg), Zn/SiO2 (ZnSi), ZnO, and MgSi. Inset: UV-
Vis spectra of different loadings of Zn on MgO/SiO2 catalysts.
(a) (b)
218
is present on all supported Cu samples, while the peak at 270 nm is present only on a Mg-
containing support. The former represents LMCT peaks for a very isolated Cu-O
species,28,44 while the latter has been assigned to an oligomeric Cu-O species.44
Analogously, the peak at 305 nm for CuSi is also assigned to the oligomeric Cu-O
species.28 This reference sample (CuSi, Figure 6.7a) also exhibits d-d transition peak at
~760 nm, indicative of Cu2+ species in a (distorted) octahedral field.28 On the other hand,
the CuMg reference exhibited an extra peak at 215 nm, possibly due to charge transfer
between Mg2+ to silica surface.27 This peak is not present on the CuMgSi catalyst which
means that Cu promotion eliminated this exposed Mg species, consistent with DRIFTS
observations. The CuMgSi catalyst exhibits a small peak at ~570 nm, which, as in the CuO
reference case, might be due to the presence of a reduced species, i.e. Cu2O or Cu0. Lower
geometry species are hardly encountered on mixed metal oxide and dehydration under inert
atmosphere is more likely to induce partial reduction on the catalyst.28 In agreement, a
known adsorption peak in the 560-570 nm region is due to the plasmon resonance of
metallic Cu nanoparticles.57
Tauc plots of CuO standard and the catalyst (CuMgSi) were derived from the UV-
Vis DRS spectra, shown in Figure S6.4. Using the method previously described by Bravo-
Suarez, et al.44, identification of the oligomer is made possible by correlating the number
of species to the edge energy. The plot in CuMgSi was deconvoluted into two species,
isolated (0 nearest neighbors) and the oligomer that will be determined, with edge energies
of 3.86 and 3.51 eV, respectively. The Tauc plot indicates that the reference oxide CuO
exhibits an edge energy of 1.26 eV, close to the previously determined values at 1.17 ±
0.06 eV.44,58 The value for the isolated CuO was higher than that reported for CuMgAl
219
mixed oxide, which had been reported to be ~3 eV.44 The difference originates from the
coordination of the isolated CuO. In the previous report it was determined that the Cu
species is in the Cu2Al domains, instead of forming a solid solution with Mg.44 Using
isolated CuO species and standard CuO (6 nearest neighbors), the coordination number, i.e.
number of Cu-O-Cu bond, was determined to be 0.8.
The Zn-promoted catalyst UV-Vis DRS spectra are shown in comparison with the
reference samples, i.e., bulk ZnO, MgSi, ZnSi and ZnMg, shown in Figure 6.7b. The
ZnMgSi catalyst shows a small peak at 276 nm. This small peak is down shifted ~100 nm,
when compared to bulk ZnO at 360 nm. Additionally, ZnMgSi contains a peak at 215 nm,
which resembles that of the CuMg UV-Vis DRS spectrum. This CT peak appears in almost
all Mg containing samples, except for CuMgSi. That peak was located at almost the same
wavelength, ~215 nm, for CuMg, ZnMg, and ZnMgSi, but shifted when MgSi support was
used, i.e. at 225 nm. This peak can be assigned to a charge transfer from Mg2+ to O2-, where
a shift is expected when MgO is wet-kneaded with SiO2.59 However, introducing Zn to the
MgSi support seems to negate this shift and it reverts back to ~215 nm. This phenomenon
is consistent with DRIFTS data, as shown in Figure 6.5, where the OH peak at 3740 cm-1
disappeared when MgO was wet-kneaded to SiO2, but reappeared when Zn is introduced
to the surface. Figure 6.7b inset shows different Zn loadings on the wet-kneaded MgSi.
At a higher loading, the peak at lower wavenumber, i.e. 215 nm, persists, while the ZnO
peak started appearing at 270 and 280 nm for 10% and 15% Zn loadings, respectively. The
shift in the CT peak is also followed by the shift in the edge energy cutoff. This shift with
a higher Zn loading was also observed by Yoshida et al. on an SiO2 support, although they
220
describe this Zn site to have a distinct electronic structure from bulk ZnO, with XANES
confirming that the ZnO is in a tetrahedral configuration.60
The reference ZnMg and ZnSi samples further aided in peak assignments of the
UV-Vis spectra of the ZnMgSi catalyst. In addition to the discussed 215 nm peak, the
former exhibits two other peaks at 276 and 360 nm. The first peak could be associated with
the defected Mg site of the catalyst, assigned to tri-coordinated O2- ions on corner sites,
which is also encountered in the MgSi sample.27,59,61 Along with the peak at lower
wavelengths, 215-225 nm, these peaks are indicative of the bulk MgO, also observed by
Figure 6.8. Scanning Transmission Electron Microscopy images of ZnMg, ZnMgSi,
CuMg and CuMgSi samples. Energy Dispersive Spectroscopy profiles (smoothed)
are also provided. Small ZnO nanoparticles are shown in ZnMgSi with red arrows.
221
Sels and coworkers.27 The second peak is likely to be assigned to bulk ZnO as a compared
to the bulk ZnO reference spectra. The ZnMgSi catalyst, on the other hand, hardly shows
any other peaks related to Zn-containing species, which rules out the bulk ZnO from active
site consideration. Bulk ZnO nanoparticles might only be formed at a very small particle
size. This is supported by STEM measurements in Figure 6.8. Chouillet et al. reported a
similar observation, where UV-Vis shows bands of a bulk ZnO phase in the limit of 1.4-
4.4 nm particle size, confirmed by TEM.47 Highly dispersed ZnO nanoparticles have also
been previously observed on MgO-supported catalysts.48,62 Another possibility is that Zn
might be present in a solid solution inside the lattice of the support (vide infra), as
previously reported in SiO247,60 or in talc.13 In particular, ZnMg shows large ~30 nm
isolated ZnO crystals present. However, ZnMgSi shows very small ~1 nm crystals and the
presence of isolated ZnO nanoparticles. This is consistent with the UV-Vis data shown in
Figure 6.7. Isolated (monomeric) Cu sites, as well as oligomeric sites in both CuMg and
CuMgSi, can’t be detected using STEM/EDS in Figure 6.8, indicating high dispersion of
these sites.
To confirm the presence of some reduced species on the surface, oxidative
treatment was done post-inert treatment by flowing air (Figure S6.5). The significant
increase in the CT bands at 250 and 310 nm in expense of the peaks at 575 and 633 nm for
CuMgSi indicates the presence of some native reduced species that became oxidized upon
the introduction of air at higher temperature. Similarly, ZnMgSi shows the continuous
increase in peaks at 230 and 340 nm, indicating the formation of both MgSi sites and bulk
ZnO phases when oxidized.
222
3. 2 Steady state catalytic performance and acid/base chemistry of the catalyst
active sites
The steady state reactivity comparison between MgSi, ZnMgSi and CuMgSi
catalysts is shown in Figure 6.9. Here the activity of three catalysts is compared in the
temperature range of 350-450°C. It can be seen that promotion with Cu and Zn significantly
enhanced the 1,3-BD formation rate from <1 mmol/gcat h to ~2 mmol/gcat h throughout the
investigated temperature range. Furthermore, ethylene formation was suppressed, more
significantly in the case of Zn promotion. The origin of this promotional effect can be
traced back to the production of acetaldehyde, which significantly increased in comparison
to the unpromoted catalyst. This accumulation of acetaldehyde on the surface indicates
that the Rate Determining Step (RDS) shifted for the case of promoted MgO/SiO2.
Quantitatively, this is confirmed by the decrease in apparent activation energy, Ea, as
derived from the Arrhenius plot of each product formation rates. Acetaldehyde and 1,3-
BD activation energy exhibits similar trend with promotion with Cu and Zn, with Ea (Zn)
< Ea (Cu) < Ea (unpromoted). Apparent activation energy of ethylene, on the other hand,
decreases with Cu promotion but not with Zn. With Zn, however, increasing temperature
does not increase the formation rate of ethylene, which explains very low activation energy
on this catalyst. The very low formation rate of ethylene must be due to very low rate
constant of ethylene formation, since raising the reaction temperature does not have
significant effect on the formation rate.
A similar increase in 1,3-BD production was previously reported by various
investigators.3 For instance, Weckhuysen and coworkers noticed a sharp increase (~20%)
in both ethanol conversion and 1,3-BD yield upon promoting the wet-kneaded catalyst with
223
1% CuO. The productivity of their catalyst was very similar to that reported here: 0.48
mmol gcat-1 hr-1 at 425°C and WHSV = 1.1 hr-1.14 When the reaction was carried out at
more than 375 °C, the conversion over ZnMgSi approached 100%. This increase in
conversion was previously observed when Zn was shown to provide more Lewis acidity
and also suppressed the Brønsted acidity.15,63 Zn-promoted catalysts, such as MgO/SiO215
and talc13, were reported to increase both the conversion and selectivity toward 1,3-BD.
The latter showed the same productivity as our catalyst, ~1.1 mmol gcat-1 hr-1 at an even
lower reaction temperature (300°C) and a much higher WHSV (8.4 hr-1).
The change in the surface chemistry of the catalyst induced by the presence of these
metals, to the best of our knowledge, has not been thoroughly investigated. The general
consensus is that the catalyst should have all redox, basic, and acidic sites on its surface.
On Cu, extensive study on the local coordination of Cu by means of XAS was not
accompanied by the identification of molecular coordination by other spectroscopic
methods.28 Further, promotional effects on Zn-promoted MgO/SiO2 catalyst were not
extensively investigated, i.e. studies were only focused on the activity change and the
implication on acid-base characteristics of the catalysts.15 Basic site poisoning using CO2
and propionic acid can reveal the reactive site difference between the three catalysts. CO2,
a relatively weaker acid than propionic acid, will occupy the stronger basic sites2,64 while
propionic acid should non-discriminatively adsorb on all basic sites given its stronger
acidity. Coflowing CO2 with ethanol as a weak acid will mainly poison the strong basic
sites and suppress any reactions that require participation of these sites. Propionic acid,
being a stronger acid, will indifferently poison any basic sites, possibly suppressing all
detectable reaction products. When switching to reactant-only flow, the weak bond
224
established between the weak CO2 molecules should be broken and therefore should
eliminate the poisoning effect and revert the system back to its original state. For propionic
acid, however, the strong basic sites should maintain strong bond with the probe molecule
after flow is stopped and should irreversibly deteriorate the productivity of one or more
reaction products that depend on the site availability.
Figure 6.9. Productivity comparison of 1,3-BD (■), ethylene (●), and acetaldehyde (▲)
over (a) MgSi, (b) CuMgSi, and (c) ZnMgSi. Dotted lines are meant to guide the eyes.
Insets: Arrhenius plots to show apparent activation energies of the three (by)products.
Reactions are carried out between 325 - 450°C, mcat = 0.1 g, pethanol = 1.8 kPa, total flow
= 55 ml/min.
225
Fundamental acid-base study on both transition metal-promoted catalysts were
investigated by both in-situ and ex-situ methods (Section S6.2). In-situ studies using
propionic acid showed that all three catalysts possessed very limited amount of strong basic
sites and that promotion with transition metals further decreased the amount of strong basic
sites. The propionic acid cofeeding experiment showed that 1,3-BD productivity did not
recover to its original formation rate which suggests the presence of some strong basic sites
that maintain strong interaction with the leftover propionic acid.2 With the wet kneaded
support, the strong basic sites are limited and more medium basic sites are present. Both
in-situ CO2 poisoning and DRIFTS study confirmed the increased availability of the
medium and weak basic sites. Our study aligns well with previous study using deuterated
chloroform, with Cu-Mg solid solution being thought of as the origin of reduced strong
basic sites.28 The in-situ poisoning further unraveled the site requirements for every step of
the reaction, i.e. acetaldehyde formation on weak basic sites, dehydration on any sites, aldol
condensation and MPV reduction on strong basic sites. The reduced amount of strong basic
sites is also the origin of RDS shift from acetaldehyde formation to MPV reduction. Total
amount of acid sites were also reduced by promotion with Zn and Cu, as shown by both
in-situ NH3 poisoning and NH3-DRIFTS experiment. While acid sites are responsible for
the dehydration steps, the origin of acetaldehyde formation rate reduction is the competitive
bonding between the available Cu2+ to NH3, since Cu catalysts are routinely investigated
as SCR catalysts.65,66 This is further supported by the recovered acetaldehyde production.
The acetaldehyde production was accompanied by Cu2+ successive reduction to Cu0, as
shown by in-situ XANES (vide infra) and was possibly the reason its productivity
decreased overtime. Promotion with transition metals yielded similar results, where Lewis
226
acid sites associated with Mg3C2+
is removed, with enhancements on the M4C2+ sites. This
finding indirectly confirms the structural change in the catalyst itself, i.e. solid solution
formation.
3. 3 Active sites under operating conditions
3. 3. 1. Temperature programmed infrared spectroscopy measurements (TP-
DRIFTS)
The effect of metal promoters on ethanol to 1,3-BD reaction mechanism were
probed using in-situ temperature programmed DRIFTS. This allowed to study the surface
species participating during the reaction. Detailed assignments of the IR peaks can be found
elsewhere.11 Briefly, experiments utilizing different probe molecules, i.e. ethanol,
acetaldehyde, crotonaldehyde, and crotyl alcohol, were performed. Table 6.4 summarized
the peak assignments from experiments done on MgSi catalyst. The in-situ DRIFT spectra
in 1700 to 1300 cm-1 region of MgSi, ZnMgSi and CuMgSi catalysts are shown in Figure
6.10 (insets). There were two very prominent peaks in the spectra at high reaction
temperatures (>250°C), i.e. ~1575 cm-1 and 1440 cm-1, previously assigned to the product
of acetaldehyde aldol condensation and polymerization.11 Noticeable difference between
the unpromoted spectra and the promoted ones was in the exact position of the two peaks.
On promoted catalysts, the C=C stretch shifted to 1587 cm-1 while the prominent peak for
the C-H bending was at 1458 cm-1. The 1587 cm-1 peak location is identical in the case for
both CuMgSi and ZnMgSi, which indicates similar anchoring site on the catalyst. As will
be discussed later some of the magnesium forms solid solution with both Cu and Zn, which
is possibly the binding site of the reaction product, given the identical peak location.
227
The C-H bending peak was very complex since every reactive intermediate has a
C-H group. Peaks were deconvoluted using CasaXPS software suite version 2.3.18PR1.167
into several different components. On the unpromoted catalysts, this broad envelope was
deconvoluted into four peaks, i.e. 1458, 1440, 1416, and 1398 cm-1. On metal-promoted
catalysts, these peaks were less convoluted showing fewer species involved with only three
prominent peaks existing. Interestingly, the peak at 1458 cm-1 was formed more rapidly in
the case of promoted catalysts, while peaks at 1435 and 1416 cm-1 lagged, compared to the
unpromoted catalyst. The growth of the peak at 1458 cm-1, previously assigned to
acetaldehyde (δ CH3) and crotonaldehyde (ρw CH3), is significantly enhanced over
promoted catalysts. The reactive nature of acetaldehyde, which is the generally accepted
Figure 6.10 Evolution of each peak during in-situ temperature-programmed ethanol
DRIFTS over (a) MgSi, (b) CuMgSi, (c) ZnMgSi. Insets: original spectra of ethanol
DRIFTS from where the peaks were deconvoluted.
228
first reactive intermediate, complicates analysis where multiple competing reactions, such
as aldol condensation, acetate formation, and polymerization to take place at low to
intermediate temperature.68–73 The adsorbed acetate formation can’t be fully ruled out due
to the peaks at 1587-1575 cm-1 that appeared and grew almost at the same rate with the
~1400 cm-1 peak.70 Together with polymerization of acetaldehyde and consecutive aldol
condensation to C6 aldehydes, these reactions present side reactions that occur. The acetate
formation is doubtful to take place in this experiment. In particular, if peak at 1587 (1575)
cm-1 is assigned to the surface acetate the change in the growth after promotion with Zn
(Cu) would apply to all the peaks in the 1460-1400 cm-1 region. In fact, the improvement
in growth of peak at 1458 cm-1 after promotion is much more significant than that for the
peak at 1587 (1575) cm-1.
Hence, the peaks at 1587-1575 cm-1 and 1457 cm-1 can be used to characterize the
degree of both aldol condensation and dehydrogenation that takes place on the surface,
while the other peaks at ~1400 cm-1 to characterize the catalysts’ basicity, i.e. its ability to
readily polymerize the formed acetaldehyde. The resulting crotonaldehyde tends to stay
on the surface and further undergo other reaction than to desorb as vapor-phase
crotonaldehyde. The C4 intermediate can be further aldolized with acetaldehyde to form
2,4-hexadienal and stick on the surface and possibly deactivate the catalyst.74 This insight
can be further utilized to probe the abundance of the active sites of the catalyst, i.e. based
on the accumulated 2,4-hexadienal which was characterized by the 1587 cm-1 peak. We
carried out semi-quantificative analysis of the peaks at 1587 (1575), 1440, and 1458 cm-1.
The peaks at ~1400 cm-1 are summed together assuming that they result from similar class
of reaction, i.e. polymerization that typically yield more than one product such as
229
metaldehyde and paraldehyde.74 The evolution of these peaks as a function of temperature
was plotted in Figure 6.10. It can be seen that for all catalysts, there was no significant
changes in the ~1400 cm-1 peak area. However, the promoted catalysts resulted in a higher
intensity/area of the 1587 cm-1 peak with Cu higher than Zn. This indicates that promoting
the catalyst with transition metal promoters enhances the ability of the catalyst to carry out
aldol condensation, while at the same time keeping the unwanted polymerization constant
with regards to the unpromoted catalyst. Another noticeable difference was the temperature
where the peak started increasing in intensity. For Cu, the peak starts increasing at lower
temperature, even at ~150 °C, while Zn lagged behind and showed similar reactivity to the
unpromoted catalyst.
Overall, combination of both DRIFTS and steady state fixed-bed experiments
showed a shift in the rate-limiting step. Without the promotion with transition metal, less
acetaldehyde was produced in the product stream indicating the rapid consumption of the
intermediate. Promoted catalysts, on the other hand, saw increase in acetaldehyde
production, which suggested a bottleneck reaction. The accumulation of acetaldehyde in
the steady-state reaction experiments suggested that aldol condensation is the RDS. The
acidity and basicity of the catalyst was affected as well by promotion with transition metal.
In-situ poisoning experiment with propionic acid and NH3 showed that promoting increases
the availability of the weak basic sites and total acid sites, as shown by the significant
decrease in the production of all products during the coflow. In-situ DRIFTS of ethanol
over the three investigated catalysts indicated that there was a change in the binding site
during the aldol condensation, as manifested by the shift of C=C stretch peak at 1575 to
1587 cm-1. This systematic change suggested that while the anchoring site was identical
230
between the two promoted catalysts, a potential solid solution formation took place.
Mechanistically, this semi-quantification confirms the steady-state experiment findings
where the activation energy of the dehydrogenation step was significantly reduced leading
to higher amount of acetaldehyde and products of aldol condensation. The change in the
polymerization products was also an indication to the altered basicity of the catalyst.68,69
Though the difference was not significant, the reduced polymerization products indicated
that the basicity of the catalyst was slightly reduced.
Table 6.4. Vibrational frequencies in 1600-1400 cm-1 wavenumber range and their
assignments for ethanol, acetaldehyde, crotonaldehyde and crotyl alcohol adsorption on
WK (1:1)11
Assignment
Experimental (cm-1)
Ethanol Acetaldehyde Enolate Crotonaldehyde Crotyl
alcohol
ν (C=C) - - 1600,
1578 1600, 1574 1602
δ (CH2) 1454 - - - 1380
δ (CH3) 1418 - - 1456, 1434 1368
ρw (CH) 1380 - - - -
ρw (CH2) - - - - 1441
ρw (CH3) 1338 1456, 1434,
1382 - 1346 1456
3. 3. 2. In-situ UV-Vis DRS study of MgSi catalysts
Figure 6.11 shows the in-situ UV-Vis DR spectra during ethanol conversion to 1,3-
BD on (a) CuMgSi and (b) ZnMgSi. The spectra plotted are difference spectra referenced
to 100 °C to better describe the dynamic changes. On CuMgSi it can be seen that with the
reaction progressing there were four broad spectral bands. Increasing the temperature lead
to the intensity increase at 248, 315 and 565 nm while the band at 276 nm showed decrease
in intensity. Interestingly, inset in Figure 6.11a shows that the band at 211 nm reached a
231
maximum at 300°C and decreased in intensity at higher temperature. To assist the peak
assignments, we performed similar experiments on unpromoted MgO/SiO2 catalyst
(Figure S6.12). The UV-Vis spectra of the unpromoted catalysts showed changes on three
bands at 210, 245, and 300 nm. These three peaks can be assigned to either CT bands of
metal oxides, π- π* transitions of allylic cations, cyclic or aromatic species, or even neutral,
uncharged aromatic species (for shorter wavelengths).75,76 An alternative assignment for
the two bands at 210 and 245 nm was the LMCT band of Mg to O on defect sites and to
SiO2, respectively.27,59 The remaining peaks are at 276 nm that decreased at the expense
of peak at 565 nm. The former was assigned to oligomeric CuO species (~0.8 Cu nearest
neighbor), while the latter one was assignable to surface plasmon resonance of Cu also due
to the rare occurrence of lower geometry CuOx species in mixed oxide systems.28,44 The
indicated reduced CuO oligomeric species to surface Cu0 will later be confirmed by X-ray
methods since peak at 565 nm could also originate from substituted or unsubstituted
benzene.75
On ZnMgSi, in-situ UV-Vis experiments showed the emergence of different
intermediates as signified the by the bands at ~240 -shifted to 268 nm at higher
(a) (b)
Figure 6.11. In-situ UV-Vis DRS under constant ethanol flow over (a) CuMgSi and (b)
ZnMgSi
232
temperature-, 300 and 211 nm. These bands were observed on the in-situ ethanol
experiment over unpromoted MgO/SiO2 catalyst as well. Another band with a cutoff at 350
nm also appeared. This band could be indicative of π- π* transitions of dienic allylic
cations75 or bulk ZnO formation since its emergence was also accompanied by the intensity
increase of shoulder at ~230 nm, which alternatively can be assigned to CT between Mg2+
to SiO2.27 The alternative assignment can suggest that Zn was transformed from its solid
solution state into bulk ZnO, as accompanied by the formation of CT band at ~230 nm.
3. 3. 3. Operando XAS studies of Cu, Zn-promoted MgSi catalysts
3.3.3.1. Operando XANES and EXAFS of Cu-promoted MgSi catalyst
The XANES spectra of Cu catalysts and standards taken under ambient condition
are shown in Figure 6.12. The XANES spectra for samples with Cu-promoted supports,
i.e. CuMg, CuSi, and CuMgSi, show similar features with a weak pre-edge peak located at
about 8977 eV and a shoulder peak at the rising edge at about 8987 eV (Figure 6.12, left).
The weak feature at 8977 eV was previously assigned to the 1s → 3d transition, and is
Figure 6.12. Normalized XANES spectra of CuMg, CuSi, and CuMgSi (a) and Cu foil,
CuO, Cu2O, and CuMg (b). Inset: Cu K-edge k2-weighted EXAFS data of
corresponding spectra. XANES spectra in Figure 6.12(a) were offset vertically for
clarity.
233
considered a signature for Cu2+ species.28,77,78 For comparison, XANES spectra of the
standards, i.e. Cu foil, Cu2O, and CuO, are plotted along with CuMg XANES spectrum.
The CuMgSi catalyst XANES spectrum strongly resembles that of the CuMg, and is very
different from CuSi and Cu standards. Further, the EXAFS spectra in the inset are very
similar for both CuMg and CuMgSi. The shoulder peak at 8987 eV, when compared to
CuO, was shifted from 8985 eV. This shoulder peak is usually assigned to the 1s → 4p
transition, and its position is associated with neighboring atomic geometry.79 For CuMg,
the shift in the shoulder peak was also observed. 28 Many reports attributed that shift to Cu
being in octahedral or distorted octahedral geometry, occupying Mg lattice sites in a solid
solution.34,35,49
Table 6.5. Best fitting results of Cu catalysts. The structural parameters of standards were
listed for comparison.
Sample Bond N R (Å)
CuMgSi Cu-O 5.6±1.1 1.96±0.02
Cu-Mg 7.0±1.8 3.01±0.02
CuMg Cu-O 4.5±0.9 1.97±0.02
Cu-Mg 7.1±2.0 3.00±0.03
CuO
Cu-O 4 1.96
Cu-O 2 2.78
Cu-Cu 4 2.9
Cu-Cu 4 3.08
Cu-Cu 2 3.18
Cu2O Cu-O 2 1.84
Cu-Cu 12 3.01
MgO Mg-O 6 2.11
Mg-Mg 12 2.98
Cu foil Cu-Cu 12 2.56
As shown in Figure S6.13 (the Fourier transformed k2χ(k) spectra of CuMgSi,
Cu2O, CuO and Cu foil), the R-space EXAFS spectra of CuMgSi have two distinct peaks
234
in the range of 1-3 Å. The peak at about 1.5 Å is due to Cu-O contribution, and the peak at
about 2.6 Å could be due to Cu-Cu contribution from Cu oxides or Cu-Mg contribution if
Cu enters MgO lattice. To determine the local environment of Cu, EXAFS analysis was
performed and two models were tested. Model A includes Cu-O and Cu-Cu path and Model
B includes Cu-O and Cu-Mg path. The fitting k range is 2.0-11.0 Å-1 and R range is 1.0-
3.1 Å. The best fitting results were obtained by using Model B and are shown in Table 6.5.
For comparison, the structural parameters for Cu foil, CuO, Cu2O, and MgO were also
listed in Table 6.5. The Cu-O bond parameters on both samples are similar to those of Cu-
O bond in the CuO. The Cu-Mg bond lengths in both CuMg and CuMgSi are also similar
to the Mg-Mg and Cu-Cu bond lengths of MgO and CuO standards, respectively. The Cu-
Cu contribution was not detected for either CuMg or CuMgSi, which corroborates the
insertion of Cu into MgO lattice. Coordination number of Cu-O shown in the EXAFS
analysis was also in line with the (distorted) octahedral geometry. Previous investigations
by Asakura et al. and Angelici et al. demonstrated that Cu-O coordination numbers were
lower than 6.28,34 Angelici, et al. found a coordination number of 4 and further assumed the
presence of two additional oxygen atoms to simulate the XANES spectra which revealed
another contribution from Cu-O bond at ~2.40 Å, which is characteristic of a separation
between copper and apical oxygen atom in a CuO6 complex.28 For CuMg, the Cu-O
contribution follows similar observation of Angelici et al. and Asakura et al., i.e. less than
6.28,34
235
Operando XAS experiments with flowing ethanol over CuMgSi were performed at
different reaction temperatures to analyze the role of Cu species during the reaction, and at
400°C, multiple scans were performed to investigate the evolution of Cu species as the
reaction progresses at constant temperature. Figure 6.13 shows the XAS spectra of
CuMgSi under both helium flow (a) and constant ethanol flow (b) at different temperatures.
As shown in Figure 6.13, the pre-edge peak (at 8977 eV), which is a signature of Cu
divalent species, remains almost unchanged after pretreatment, indicating Cu remains in
the 2+ state after He treatment. Under helium at elevated temperatures, a new feature at
8982 eV appeared suggesting the change of the local environment of Cu after pretreatment.
The position (8982 eV) of this peak is quite close to that (8981 eV) of the shoulder peak of
Cu2O, in which each Cu atom is surrounded by two O atoms in a collinear manner. The
appearance of the 8982 eV peak thus implies the decrease of the average coordination
number of Cu-O bond for Cu atoms in CuMgSi catalyst. During experiment with ethanol,
significant increase in the intensity of 8982 eV peak was observed especially at high
Figure 6.13 Normalized temperature-programmed operando XANES spectra of
CuMgSi catalyst under He flow (a) and ethanol flow (b). Inset: enlarged region of the
pre-edge features to elucidate changes at different temperatures.
236
temperatures, suggesting the increased fraction of species in which the average Cu-O
coordination number is low. We propose that such geometry is correlated with catalytic
activity of CuMgSi catalyst. The corresponding MS data (Figure S6.14), shows that the
acetaldehyde was made at very low temperature, i.e. starting as low as 100 °C, and
increased significantly at ~250 °C. This increase is reflected as well by the spectra at 300 °C,
where the increase is very significant from 200 °C. At the same time, the 1,3-BD started
being produced at ~250 °C, which was lower than unpromoted catalyst, i.e. 300 °C.
When reaction temperature reached 400 °C, the temperature was held constant
while XANES spectra were repeatedly taken to investigate any changes that take place
during the reaction. The change in the copper species was recorded as a function of time,
shown in Figure 6.14. A Cu foil XANES spectrum taken at ambient temperature was
overlaid for comparison. As reaction proceeded, the peak at 8982 eV started decreasing in
intensity, suggesting the re-arrangement of the local structure of Cu. Accompanied with
Figure 6.14. Normalized time-resolved operando XANES spectra of CuMgSi catalyst
under ethanol flow at 400°C. Inset: enlarged region of the pre-edge features to elucidate
changes at different temperatures.
237
this decrease, the peak at 8980 eV which is also a feature of Cu foil spectrum showed up
and increased with time, suggesting the formation of Cu metallic phase. Basing on the
above results, we conclude that changes in the local structure of Cu occurred throughout
the reaction. The quantitative information on the local structure of Cu during the reaction
conditions was obtained by performing EXAFS analysis and the results were summarized
in Figure 6.15. Figure 6.15 shows the change in the coordination numbers of Cu-Cu, Cu-
Mg, and Cu-O bonds during the reaction. From 200-400 °C, a steady decrease in Cu-O
bond coordination number takes place, which, as discussed above, is also manifested by
the increase in the intensity of 8982 eV peak. There was no appearance of Cu-Cu bond
until the steady-state condition at 400 °C. At 400 °C, the final EXAFS spectra show a
significant increase of Cu-Cu coordination number from 0 to about 3. This indicates
clustering of the Cu atoms after reaction has stabilized at 400 °C.
To confirm the correlation between the XANES features with the coordination
number of Cu-O bond, XANES spectra simulations were performed using FEFF 9 code.80
Figure 6.15. Coordination number changes during reaction of ethanol to 1,3-BD over
CuMgSi
238
Simulations were first performed on CuO and Cu2O to find optimized simulation
parameters, which were then applied in calculating the spectra of all models. For the as-
prepared CuMgSi catalyst, according to EXAFS analysis, the coordination number of Cu-
O was close to 6 and Cu is very likely taking the Mg sites in MgO lattice. We therefore
built an MgO sphere which contains 251 atoms and has diameter of about 1.6 nm, and
replaced the core Mg atom by a Cu atom. This model was named Model 1. In this model,
Cu is octahedrally coordinated by 6 O atoms at the same distance. The calculated XANES
spectrum of this model is plotted in Figure 6.16, and the shoulder peak at the rising edge
is indeed shifted to higher energy compared to that of CuO, which agrees with the trend
observed in experimental data. As shown by EXAFS results, under reaction conditions and
at high temperatures, the average Cu-O decrease and is close to 4. We thus modified Model
1 by removing 2 oxygen atoms around Cu. In this modified model, Model 2, Cu is then
surrounded by 4 oxygen atoms at the same distance forming a planar geometry. In the
simulated XANES spectrum of Model 2, a shoulder peak appears in position between those
of Cu2O and CuO. Such trend was also observed in the experimental spectra. Therefore,
the agreement between the experimental and theoretical XANES spectra suggests the
shoulder peak at the rising edge of Cu spectra is related to the local oxygen environment
around Cu. In the CuMgSi system, Cu replaces Mg in MgO lattice. When the reaction
occurs, the octahedral Cu-O geometry will be distorted: most likely, part of oxygen atoms
are pulling away from Cu, which could be then transformed to Cu metallic phase as
detected in the final aged catalyst (Figure 6.14).
239
A complementary view of this operando measurement was offered by Angelici et
al., where reactions were carried out at 400 °C under two different pretreatment conditions,
i.e. inert flow and reducing atmosphere.28 Under inert flow, the initial state of the catalyst
consists of the native distorted octahedral Cu2+ species that was originally in the catalyst
and another Cu2+ species that resembles to Cu2+ from CuO/SiO2. This latter Cu2+ species
was reduced to Cu0 and transformed to the distorted octahedral Cu2+ species when
pretreated at 425 °C under inert flow. Our observations show that there are new Cu species
as evident by the peak at 8982 eV that appeared when catalyst was pretreated at high
temperature even though the pre-edge feature at 8977 eV, assigned to the distorted
octahedral Cu2+ from CuMgSi, barely changed. Interestingly, similar distribution between
Cu2+, Cu+, and Cu0 was observed after ethanol reaction without reducing pretreatment, after
Figure 6.16. XANES spectra of the simulated CuO Model 1: Cu in a local environment
surrounded by 6 oxygen atoms and Model 2: Cu in a local environment surrounded by
4 oxygen atoms.
240
reducing pretreatment under H2 and after ethanol reaction with reducing pretreatment.28
Specifically, the three steps mentioned correspond to increasing amount of Cu0 in the final
state of the catalyst. This indicates that both ethanol and hydrogen have a competing
reducing effect on the catalyst. The final state after the steady-state reaction under both
pretreatment conditions revealed that there were some Cu2+ species on the catalyst even
after extensive reaction with ethanol.28 In our experiments, however, we observed a
different outcome. The two pre-edge features at 8977 and 8987 eV behaved similarly with
both of them barely changing during the reaction. Even after extensive reaction at 400°C,
Cu-Mg coordination number did not change, while Cu-O coordination number decreased
(Figure 6.15) to 4. The apparent increase in peak at 8987 eV is mostly due to the increase
in peak at 8982 eV. We propose, based on data in Figure 6.13-15, that origin of the peak
at 8982 eV, assigned to Cu2+ with less-than-6 oxygen neighbors, is from a bulk Cu2+ with
six oxygen neighbors that catalyzed the reduction and lost bonding with two neighbor
oxygens during interaction with ethanol, as indicated by the simulation (Figure 6.16).
Furthermore, this new Cu species undergoes change in coordination number, decreasing to
reduced Cu0, possibly due to the depleted reducible Cu2+ that shifts the reaction active sites
and further reduced all reducible copper species into Cu0, as suggested by clustering of Cu
(increase in Cu-Cu coordination number) as the reaction progressed at 400°C. The
decreased reducibility of Cu2+, evident from the presence of Cu2+ at the end of the reaction,
was also observed previously on CuZn catalysts supported on MCM-41 and Al2O3, where
co-presence of Zn2+ led to the formation of isolated Cu2+ species that was reduced at higher
temperature.42,81 Other factors that deteriorate Cu2+ reducibility can be attributed to the
presence of solid solution phase and bulk CuO phase, such as that found in CuMnZrO2 and
241
CuMgAlOx hydrotalcite catalysts, respectively.44,78 The operando XANES and in-situ UV-
Vis confirmed the presence of two Cu species on the catalyst prior to exposure to ethanol,
i.e. distorted octahedral Cu2+ (possibly from solid solution) and reducible Cu2+ species, as
suggested by in-situ dehydrated UV-Vis as well.
3.3.3.2. Operando XANES and EXAFS of ethanol over Zn-promoted MgSi catalyst
The XANES spectra of Zn catalysts and standards taken in ambient condition are
shown in Figure 6.17a. The standards used in this study are Zn foil and ZnO to represent
the reduced and oxidized states of the transition metal. Comparison between ZnMgSi, ZnSi
(ZnO/SiO2), and ZnMg (ZnO/MgO) reveals similarity between ZnMgSi and ZnMg. The
silica-supported sample looks like those of willemite or hemimorphite, both Zn-silicates.47
Chouillet et al. investigated the effect of drying temperature prior to calcination, and
XANES spectra of all dried samples calcined at 450 °C, only 50 °C lower than our
temperature, are nearly identical and indicative of zinc silicate formation.47 The Zn foil
exhibits a peak at 9660 eV, which was assigned to electron transition to empty d orbital.
The absence of this feature indicates that all samples are fully oxidized.55 For Zn standards
(ZnO and Zn foil), there are two main features, the main edge, labeled as A, and feature B
in the spectra. The main peak was assigned to 1s4p electron transition with lesser peak
intensity corresponding to decreasing coordination number of the cation.82–84 The second
feature was a multiple scattering resonance associated with medium range molecular
structure around the target element; this feature was located differently for each sample,
indicating difference in geometric molecular structure.82,83
242
Both Mg-containing samples, i.e. ZnMg and ZnMgSi, exhibit splitting at the edge
that was significantly larger than that of ZnSi. The splitting was previously observed on
ZnO/Al2O3 and ZnFe2O4 as well and was attributed to a Zn2+ structure in a rigid
environment nothing like ZnO.83,85 EXAFS spectra of the samples show very similar
spectral shape between the two samples although the oscillation magnitude of the ZnMgSi
sample was much lower. The similarity indicates that the Zn in both samples possess very
similar local structure. Fourier transform was applied to the EXAFS signal (k2χ(k)) of
ZnMg to represent both samples and compared to ZnO and Zn foil (Figure 6.17b).
Between 1-3 Å, there are two peaks at 1.40 Å and 2.40 Å. From the Fourier transformed
spectra the first peak was attributed to Zn-O bond, while the latter was lower than Zn-Zn
bond length in ZnO yet higher than Zn-Zn bond length in Zn foil. This implies that this
was not due to the contribution of Zn-Zn bond and we predict this to be Zn-Mg bond. To
confirm it, we did EXAFS analysis for the ZnMgSi catalyst and tested three models: Model
A includes Zn-O and Zn-Zn paths; Model B includes Zn-O, Zn-Zn, and Zn-Mg paths;
Figure 6.17. (a) Normalized XANES spectra of ZnMg, ZnSi, ZnMgSi, Zn foil, and
ZnO. Inset: Zn K-edge k2-weighted EXAFS data of corresponding spectra. (b) Fourier
transforms of the EXAFS spectra of ZnMg, ZnO, and Zn foil.
243
Model C includes Zn-O and Zn-Mg paths. The fitting k range is 2.0-10.5 Å-1, and R range
is 1.0-3.2 Å. Model 3 provides us best fitting results, which confirms that Zn was singly
distributed into MgO lattice. This Zn-Mg bond was ~0.2 Å shorter than that of Zn-Zn bond
in the ZnO foil, which was also previously determined in Zn(1-x)MgxO alloy.86 The bond
length values for standards and samples are tabulated in Table 6.6.
Figure 6.18. Normalized temperature-programmed operando XANES spectra of
ZnMgSi catalyst under He flow (a) and ethanol flow (b). Inset: enlarged region of the
pre-edge features to elucidate changes at different temperature. (c) Temperature-
induced change in coordination number of Zn-Mg and Zn-O bonds during the reaction.
(c)
244
The operando XANES spectra during ethanol conversion are presented in Figure
6.18. Similar to study on CuMgSi, the experiment was conducted with increasing
temperature under ethanol flow (Figure 6.18b). The MS data of the experiment shows
similarities with CuMgSi. In particular, acetaldehyde was produced very early as well
following the induction time between ethanol flowing into the reactor and the product
stream entering the MS. The production of 1,3-BD follows similar trend, i.e. started being
produced at lower temperature before really ramping up at ~300 °C. This sudden increase
at 300 °C coincides with the increase in acetaldehyde production as well, which suggests
that there are two active sites for ethanol dehydrogenation for both catalysts. The presence
of these two sites on two promoted catalysts indicates that there are identical sites on both
catalysts. When compared to unpromoted MgSi catalyst, the acetaldehyde production was
found to dramatically increase at this temperature as well. This indicates that Zn and Cu
both are present as an additional dehydrogenating site, and that the native weak basic sites
responsible for the reaction are still present after promotion.
The Zn2+ local structure, however, shows a resilience nature with flowing ethanol,
as shown in Figure 6.18b (inset). There was no significant change under ethanol flow,
compared to the thermal effect when only helium was flown (Figure 6.18a). Figure 6.18c
further shows the analysis of the EXAFS spectra where there was no significant changes
in Zn local coordination number (N) during the reaction. The identified Zn-Mg and Zn-O
both remained intact with no change in the local state of the catalyst was observed. This
indicates that the Zn-promoted catalyst should be relatively stable compared to Cu-
promoted catalyst and possible deactivation is more likely to be related to the formation of
245
carbonaceous deposit on the surface due to the higher activity exhibited by the additional
redox and Lewis acid sites provided by the Zn dopant.15
Table 6.6. Best fitting results for ZnMgSi, ZnMg, ZnO, MgO and Zn. The structural
parameters of standards were listed for comparison.
Sample Bond N R (Å)
ZnMgSi Zn-O 3.6±0.5 1.98±0.02
Zn-Mg 4.8±1.6 3.09±0.04
ZnMg Zn-O 4.7±1.0 2.09±0.04
Zn-Mg 14.0±2.8 3.05±0.02
ZnO
Zn-O 4 1.94
Zn-Zn 6 3.15
Zn-Zn 6 3.2
MgO Mg-O 6 2.11
Mg-Mg 12 2.98
Zn foil Zn-Zn 6 2.66
Zn-Zn 6 2.88
4. Conclusions
Cu- and Zn-promoted wet kneaded MgO/SiO2 catalysts were interrogated in situ
and operando and provided new insights into the structure and reactivity of their catalytic
sites during ethanol reaction to 1,3-BD. No distinct crystalline promoter phases were
obtained according to XRD and STEM measurements and Cu and Zn was suggested to
bind strongly with the native OH groups. Under dehydrated conditions, oligomeric Cu-O
species were found to dominate CuMgSi while the combination of very small <4 nm ZnO
nanoparticles and possibly solid Zn solution with MgO have been observed using a
combination of UV-Vis and STEM measurements. The reduced amount of strong basic
sites due to the metal promoter binding was found to affect RDS shift from acetaldehyde
formation to MPV reduction. In situ DRIFT spectroscopy results allowed to decouple the
246
aldol condensation and dehydrogenation fundamental steps that takes place on the surface
suggesting that promoting the catalyst with transition metal promoters enhanced the ability
of the catalyst to carry out aldol condensation as correlated with the steady state reactivity
experiments. In situ UV-Vis spectroscopy suggested appearance of π- π* electronic
transitions of allylic cations, cyclic or aromatic species on the catalysts while also
providing insights on the oligomeric structure of the active sites. In particular, oligomeric
CuO species with ~0.8 Cu nearest neighbor were found to decrease in intensity suggesting
their involvement in ultimate catalytic Cu0 species formation.
Our operando X-ray measurements were combined with ab initio multiple
scattering modelling to unravel the exact electronic structure of the Cu and Zn promoters.
These measurements were performed as a function of temperature and signified that Cu-
Cu bond appeared at reaction temperatures of 400 oC on the aged (TOS of 6-7 hours)
catalyst at the expense of Cu-O bonds. Cu replaced Mg in MgO lattice to eventually lead
to Cu aggregates. This is akin to the literature reports where deactivation of Cu-containing
catalysts was suggested due to the carbonaceous deposits rather than sintering of the
promoter. Furthermore, the 8982 eV peak typically assigned to Cu+ species, in our work
was assigned to a 4-fold coordinate Cu species, rather than Cu2O and is proposed as the
key intermediate leading to increase in Cu-Cu bond number. It is transient and is only
populated at temperatures lower than 400 oC and starts decreasing to yield Cu0 during aging
with ethanol. Two types of Zn bonds, namely Zn-O and Zn-Mg, were identified during X-
ray analysis and showed resilience to ethanol under operating conditions. Particularly, Zn
was nearly 6-coordinated when in the vicinity of Mg while Zn-O species showed nearly 4
nearest neighbors.
247
Chapter 7 – Supporting Information
S6.1 Catalyst Characterization
Figure S6.1. XRD patterns of (a) Zn/MgO and (b) Zn/SiO2 at different loadings.
Figure S6.2. XRD patterns of (a) Cu/MgO and (b) Cu/SiO2 at different loadings.
248
Figure S6.3. In-situ DRIFTS of OH region of dehydrated MgSi catalysts references at
100°C for Cu-promoted (left) and Zn-promoted (right). Spectra are offset for clarity.
Figure S6.4. Tauc plot of CuO (left) and deconvoluted Cu species of CuMgSi catalyst
(right) to determine the edge energy/band gap (E0) for correlation with number of Cu
coordination.
Figure S6.5. In-situ UV-Vis difference spectra of oxidative dehydration of (a)
CuMgSi and (b) ZnMgSi.
(a) (b)
249
S6.2 Catalyst Acid-Base characterization
The CO2 coflow is shown to inhibit 1,3-BD production on all catalysts while also
increasing the production of acetaldehyde, except for ZnMgSi. CO2 interacts only with
strong basic sites, give its nature as a weak acid, and the change in acetaldehyde and
Figure S6.6. Poisoning reactivity testing using CO2 to determine the role of basic sites
during ethanol conversion to 1,3-BD over (a) MgSi, (b) CuMgSi, and (c) ZnMgSi.
Reactions are carried out at 400 °C, mcat = 0.1 g, pethanol = 2.5 kPa, total flow = 55 ml/min.
All formation rates are normalized to initial 1,3-BD formation rate.
250
ethylene production suggests the role of weak basic sites and strong basic sites in catalyzing
the dehydrogenation and dehydration steps, respectively. The experiment on metal-
promoted catalysts suggest that promotion with Cu or Zn decreases the strong basic sites
since the ethylene production is not severely affected. The interaction between the catalysts’
surface with CO2 was studied by means of in-situ DRIFTS using CO2 as a probe molecule.
The experiment corroborates the in-situ poisoning experiments, where the amount of strong
basic sites, demonstrated by the peaks assigned to monodentate and polydentate carbonate,
are reduced upon introduction of transition metal promoters.87 Upon introduction of the
promoters, change in the stability of both monodentate and polydentate carbonate is
lowered at higher temperature, with carboxylate species is now formed on the catalyst. CO2
adsorption on transition metal oxides generally yields an additional carboxylate species,
which is stabilized by back-bonding between d-orbitals of the metal ion and π* orbital of
the C=O bond.88 Given its much lower stability, this carboxylate must indicate a weaker
basic site that is present on the catalysts. Peak assignments of the CO2 surface species are
tabulated in Table S1.
251
Table S6.1. Peak assignments of surface CO2 species identified on MgSi, CuMgSi, and
ZnMgSi catalysts.
Species Vibrational mode Catalysts
MgSi CuMgSi ZnMgSi
Monodentate
carbonate
υas OCO 1506 1500 1500
υs OCO 1441 1440 1440
Bidentate
carbonate
υas OCO 1663 1611 1611
υs OCO 1362 - -
Symmetrical υ OCO 1425 - -
Polydentate
Carbonate
υas OCO 1571 1573 1566
υs OCO 1380 1384 1391
Bicarbonate
υas OCO 1634 1644 1644
υs OCO 1460 1464 1464
δ OH 1280 1290 ~1290
Carboxylate υas OCO N/A 1593 1593
υs OCO N/A 1374 1374
Poisoning using propionic acid shows significant reduction in 1,3-BD production
for all catalysts. The reversibility nature of this change indicates that these catalysts in
general had very little amount of strong basic sites. The experiments demonstrated that the
acetaldehyde formation rate was more significantly affected during propionic cofeeding on
promoted catalysts, i.e. CuMgSi and ZnMgSi. From Figure S6.8, it is apparent that the
basicity was very different for the transition metal-promoted catalysts. Propionic acid was
shown to interact more strongly with the transition metal sites deactivating active sites
more readily, as shown by the degree of retardation it caused during propionic acid
cofeeding.
Coflowing NH3 resulted in poisoning of acid sites of the catalysts, and
mechanistically, these sites are responsible for the dehydration steps that follow aldol
condensation.2 In this work, ethylene and 1,3-BD production were adversely affected in
both ZnMgSi and CuMgSi, as opposed to MgSi. Which indicates the higher availability of
total acid sites on unpromoted MgSi catalyst. The decrease in acetaldehyde production on
252
CuMgSi is due to competitive adsorption and activation of NH3, since CuO catalysts are
well-known SCR catalysts.65,66 Zn promotion, on the other hand, showed a very strong
dehydrogenation enhancement with acetaldehyde being accumulated on the catalyst and
shifting the RDS to MPV reduction.
253
Figure S6.7. CO2 Temperature Programmed-DRIFTS on (a) MgSi, (b) CuMgSi, and
(c) ZnMgSi.
(b)
(a)
(c)
254
Figure S6.8. Poisoning reactivity testing using propionic acid to determine the role of
basic sites during ethanol conversion to 1,3-BD over (a) MgSi, (b) CuMgSi, and (c)
ZnMgSi. Reactions are carried out at 400 °C, mcat = 0.1 g, pethanol = 2.5 kPa, total flow
= 55 ml/min. All formation rates are normalized to initial 1,3-BD formation rate.
255
The surface acidity of the catalysts was investigated with DRIFTS using NH3 as a
probe molecule. NH3 is the most commonly used probe molecule, due to its small size,
which can penetrate all sites available on the catalyst without being limited by catalyst
Figure S6.9. Poisoning reactivity testing using NH3 to determine the role of acid sites
during ethanol conversion to 1,3-BD over (a) MgSi, (b) CuMgSi, and (c) ZnMgSi.
Reactions are carried out at 400 °C, mcat = 0.1 g, pethanol = 2.5 kPa, total flow = 30 ml/min
(without NH3), 55 ml/min (with NH3). All formation rates are normalized to initial 1,3-
BD formation rate. NH3 desorption spectra on MgSi catalysts at 100°C are shown in
(d).
(d)
256
geometry.89–91 The Lewis acid sites are discriminated by the bending and stretching modes
of the coordinated NH3, i.e. NH3 unpaired electron donated to metal sites and ammonium
ion formed due to the strong OH acid sites.90 From Figure S6.9d, peaks at 1650, 1615,
and 1590 cm-1 are found in the N-H deformation region. The absence of peak at ~1450 cm-
1 indicates the non-existence Brønsted acid sites, while the aforementioned peaks are
assigned to two different Lewis acid sites. Using the help of DFT, peaks at 1650 and 1615
cm-1 are assigned to asymmetric N-H bending mode of Lewis-bound NH3 species, while
1590 cm-1 is assigned to symmetric N-H bending mode of the Lewis-bound NH3 species.
In particular, 1650 cm-1 and 1590 cm-1 represent the same species that disappeared upon
promotion with transition metal sites; assigned to Mg2+3C from our calculation, while the
corresponding symmetric bending mode should be around ~1580 cm-1 give the split. Note
that the DRIFTS simulation of both open and closed sites does not lead to proper
discrimination of both sites, and hence this technique should not be used to differentiate
both sites (Table S2). The Several other peaks at 1490, 1400, and 1370 cm-1 are associated
with dissociative adsorption of NH3 that takes place at low temperature.89
Table S6.2. DFT simulation of NH3 on MgO slab. Simulation was done using VASP, PBE
functionals on 2x2x1 k-point mesh.
Type Binding site Vibrational mode
δas NH2 δs NH2
Open 4C 1616 1590
3C 1598 1557
Closed 4C 1612 1595
3C 1645 1601
257
To further characterize the active sites on the MgSi catalysts, operando methanol
DRIFTS-MS was carried out due to its versatility as a probe molecule.92,93 On the basic
sites methanol produces CO2, acidic sites yield dimethyl ether (DME), while redox sites
will form formaldehyde.93 Hence, this operando testing allows further measurement of
changes in catalyst redox properties, not directly available via CO2 and NH3 testing. The
DRIFTS spectra of MgSi catalysts are shown in Figure S6.10. The C-H region (Figure
S6.10, left) is typically used to identify the presence of surface methoxide.92 Upon CH3OH
adsorption on the surface, several peaks showed up at 100 °C, i.e., 2990, 2965, 2920, 2860,
2820, and 2780 cm-1. Methoxy species bounded to SiO2 sites, such as Si-OCH3, are
indicated by the peaks at 2990, 2965 and 2860 cm-1, assigned to υas (CH3), υs (CH3), and
2δs (CH3), respectively.94 The analogs of these peaks for Mg-OCH3 are located at 2920,
Figure S6.10. DRIFTS spectra in the C-H stretching (left) and bending (right) region
of methanol desorption under He flow on unpromoted (top) and promoted (bottom)
catalysts.
258
2820, and 2780 cm-1.95,96 The presence of the two kinds of adsorbed methoxy species have
previously been observed by Wachs and coworkers.94 In their study, it was shown that
surface methoxide was present as Si- and V- bounded in the case of V2O5/SiO2 catalyst. A
shoulder at 2880 cm-1 is clearly seen on Cu- and Zn-promoted samples, but is lower in
intensity for the unpromoted sample. This peak has previously been assigned to the υs (CH)
of formaldehyde, formed by the readsorption of the redox product.94 The peak at >350 °C,
i.e., 2805 cm-1, can possibly be attributed to surface formaldehyde υ (CH), based on
Busca’s work.97
Methanol typically adsorbs as two kinds of species, as surface methoxy
(dissociative adsorption) and as a molecularly bonded species to Lewis acid site in a minor
amount.94 The asymmetric and symmetric methyl bends of the former are located around
~1450 and ~1430 cm-1, while the second species is characterized by its OH bending at
~1360 cm-1 (Figure S6.10, right). These three peaks can be found on the spectra at low
temperatures, while they disappear at higher temperatures. The adsorbed methoxy species
further dehydrogenate into surface formate via C-H bond breaking on the redox site, and
the basic site will perform another C-H bond scission to make carbonate.93 The bicarbonate
species, as explained before, is characterized by peaks at 1644 and 1464 cm-1. The latter is
overlapped by the methoxide methyl bending mode, but still apparent due to the broadness
of the peak. An additional peak at 1670 cm-1 also occurred in these spectra, as it does for
CO2 adsorption, illustrated in Figure S6.10. At higher temperatures monodentate
carbonate is apparent at 1386 cm-1 and is accompanied by a shoulder at ~1587 cm-1. This
is obscured by the intensity of the bidentate carbonate peak at 1611 cm-1. The presence of
surface formate in this experiment is revealed by the peaks at 1595 and a peak at around
259
~1340 cm-1, assigned to υa (OCO) and υs (OCO), respectively.97 The prevalence of surface
formate on the unpromoted catalyst further demonstrates the basicity of the catalyst. The
peaks at 1595 and 1340 cm-1 are much less pronounced in the spectra of the Cu- and Zn-
promoted catalysts. This could indicate the spontaneous desorption of the produced
formaldehyde, even though at higher temperatures, re-adsorption of formaldehyde is more
pronounced in the case of promoted catalysts, as shown by the peak at 2880 cm-1 (Figure
S6.10, left).
The corresponding MS data from the operando methanol spectroscopy are shown
in Figure S6.11. As discussed, methanol adsorbs in two different ways, by dissociative
adsorption and by molecular adsorption on Lewis sites. This is further corroborated by the
vapor phase MS data, which shows a methanol peak (m/z = 31) for each catalyst, consisting
of two different peaks. The symmetry of these peaks indicates that they consist of two
peaks. A second peak, which is apparent as a shoulder at 300 °C, indicates the release of
two adsorbed methanol species into the vapor-phase. The lower temperature peaks for the
methanol occur at temperatures below 200 °C for each catalyst and are due to the strongly
bound, yet molecularly adsorbed, methanol species on the Lewis acid site. The higher
temperature peak is due to the recombination of the surface methoxide and surface
hydroxyl group. The most striking observation is for the formaldehyde spectra (m/z=29),
where the Tp values are situated close to their corresponding methanol Tp peaks. On redox
sites, methanol dissociates to give surface methoxide (CH3O·) and surface hydroxyl group
(OH·). The surface methoxide will then perform a subsequent C-H bond breaking step and
desorb as formaldehyde.98 The un-promoted sample shows a very close Tp for both
methanol and formaldehyde (190-193 °C), while the Cu-promoted and Zn-promoted
260
samples show a significantly lowered temperature peak at 177°C. This shows that the redox
capability of the catalyst has been improved by transition metal doping. The formaldehyde
peak at lower temperature is very close to the methanol peak, at 182 and 185 °C for ZnMgSi
and CuMgSi, respectively, which differs only by about 5-8 °C. Formaldehyde is known to
re-adsorb onto the surface94 so the second peak at a higher temperature originates from the
desorption of this formaldehyde species.
The CO2 temperature profiles (m/z=44) are also shown in Figure S6.11. The
mechanism of the methanol to CO2 reaction is complex. Surface formate is required as a
surface intermediate and this requires surface oxygen, since formaldehyde possesses only
one oxygen atom. The re-adsorption of formaldehyde into surface formate is corroborated
by the DRIFTS data shown in Figure S6.10 and induced by the presence of the basic sites
Figure S6.11. Online MS analysis during operando methanol DRIFTS of CuMgSi,
ZnMgSi, MgSi and reference MgO
261
on the surface.93 Subsequently, C-H bond breaking takes place and then CO2 is released.
Pure MgO is used in this experiment as a reference. On MgO the Tp peaks for methanol,
formaldehyde and CO2 are very close to each other, at 198, 200, and 200 °C, respectively.
This indicates a competing, consecutive reaction for both formaldehyde and CO2 formation.
In this case, the desorption of formaldehyde from the surface after methoxy C-H bond
breaking has a very similar rate with subsequent HCOO- formation and C-H bond breaking
to give off CO2. The redox capability of MgO is also acknowledged by Badlani and Wachs,
where a steady-state reaction of MgO yields both formaldehyde and CO2 with higher
selectivity towards the former.93 The presence of the second CO2 peak at 315 °C represents
the secondary formation of CO2 from the re-adsorption of formaldehyde. The CO2 peak
never fully disappears, since there is a small amount of CO2 being released from bulk
magnesium carbonate.99 Interestingly, this basicity of MgO is not reflected in the MgSi
catalyst. The CO2 peak is practically non-apparent as shown in Figure S6.11. SiO2 is a very
inert material, which should be the reason why there is very little CO2 in the vapor-phase.
This is mostly due to methanol thermal decomposition.93 Wet-kneading with SiO2 (1:1)
should reduce the number of MgO basic sites and induce the formation of different sites,
such as closed or open Lewis acid sites. These are formed from the Mg-O-Si linkages11
and additional redox sites, as shown by the increased formaldehyde production for MgSi.
The increased number of redox sites is apparently caused moreso by the higher
formaldehyde productivity than by the pure MgO, which is why most MgO catalysts won’t
perform the ethanol-to-1,3-BD reaction. From our CO2 DRIFTS experiment in Figure S6.7,
it can be seen that the CO2 peaks are present on all catalysts at elevated temperatures, with
MgSi possessing the most integrated area. Un-promoted MgSi shows a better retention of
262
CO2, demonstrated by the higher intensity of the carbonate peaks at a higher temperature,
as compared to ZnMgSi and CuMgSi (Figure S6.7). It is likely that once the C-H bond of
the surface formate is broken, the CO2 formed is bound to an oxygen site on the
unpromoted catalyst, whereas promotion with Zn or Cu reduced the electronegativity of
the nearby oxygen site and released CO2 at a higher rate than for the unpromoted sample.
A critical analysis done in this work is the comparison of the redox capability of
the catalysts, which was done by evaluating the activation energy of the methanol
dehydrogenation reaction to yield formaldehyde. The decomposition kinetics of the C-H
bond breaking of surface methoxy is known to follow a first order reaction with pre-
exponential factor of ~1013 s-1.100 The activation energy was calculated using the Redhead
equation101
𝐸𝑎
𝑅𝑇𝑝2 =
𝜐
𝛽𝑒
(−𝐸𝑎
𝑅𝑇𝑝) (1)
where Ea is the activation energy (J/mol), R is the gas constant (J/mol K), Tp is the TPSR
temperature (K), β is the heating rate (10 °C/min), and υ is the pre-exponential factor (s-1).
From equation (1), activation energies of 28.4, 27.7 and 26.7 kcal/mol for reference
MgO, MgSi and the CuMgSi and ZnMgSi were calculated, respectively. The lowered
activation energy explains the more reactive nature of the catalyst, since promotion of the
catalyst has been shown to give a lower activation energy for the dehydrogenation reaction.
Promotion with Zn and Cu enhances the alcohol dehydrogenation capability.13,14 This
promotional effect doesn’t carry over to the ethanol dehydrogenation in a straightforward
manner, since for ethanol, utilizing Zn and Cu promotion has shown a very profound effect
on the ethanol dehydrogenation step of the reaction.3 Similarly, the non-existence of DME
263
as the product of the acidic sites can’t be translated to the catalyst inability to carry out the
dehydration reaction. DME formation requires two available sites that bind two ethoxy
species, while for ethylene formation only one site is necessary. The absence of DME in
the product goes along well with our steady-state experiment, where hardly any diethyl
ether (DEE), an analog for ethanol bimolecular dehydration, is produced. The inability of
the catalyst to produce DEE does not necessarily mean it is not acidic, since ethylene is
prominently produced during the reaction. This experiment suggests that all the catalysts
used in this work do not possess two neighboring acidic sites, which are required to
dehydrate alcohols. Furthermore, the semi-quantification of the active sites for redox sites
can be calculated by integrating the area under the peak, referenced to the surface area of
the catalysts, which are calculated using BET method (Table S6.3).
Table S6.3. Redox properties of the MgSi, CuMgSi and ZnMgSi catalysts and reference
MgO obtained from MS measurements. These results have been normalized to the BET
surface area (m2/g) of each catalysts.
Catalysts Redox site density relative
to MgSi
Activation energy
(kcal/mol)
Reference MgO 0.48 28.4
MgSi 1.00 27.7
CuMgSi 0.42 26.7
ZnMgSi 0.96 26.7
Promotion of the catalyst with a transition metal unexpectedly reduced the number
of redox sites, which indicates the loss of these sites when the catalyst is promoted with
transition metals. Promotion with Cu reduced the Redox site density significantly, while
Zn barely modified the redox site density, which can be related to the superior ethanol
reactivity in the steady-state experiment. The decrease in Redox site density further
264
justifies explanations by previous reports that higher loading would be detrimental to the
catalyst activity, as well as Zn’s significant enhancement in the ethanol’s conversion.3
S6.3 In-situ and operando characterization of Cu and Zn coordination on promoted
MgSi catalysts
Figure S6.12. In-situ UV-Vis DRS of ethanol reaction on undoped MgO/SiO2 catalyst.
Difference spectra is shown, where catalyst spectra at 100°C with chemisorbed ethanol is
used as a reference.
Figure S6.13. R-space EXAFS spectra of CuMg catalyst, in comparison to Cu foil, CuO,
and Cu2O
265
Figure S6.14. Corresponding MS data of in-situ XANES-EXAFS for ethanol to 1,3-
BD over (a) CuMgSi, (b) ZnMgSi
266
References
(1) Posada, J. A.; Patel, A. D.; Roes, A.; Blok, K.; Faaij, A. P. C.; Patel, M. K.
Bioresour. Technol. 2013, 135, 490–499.
(2) Shylesh, S.; Gokhale, A. A.; Scown, C. D.; Kim, D.; Ho, C. R.; Bell, A. T.
ChemSusChem 2016, 9 (12), 1462–1472.
(3) Pomalaza, G.; Capron, M.; Ordomsky, V.; Dumeignil, F. Catalysts 2016, 6 (12),
203.
(4) Sushkevich, V. L.; Ivanova, I. I. ChemSusChem 2016, 9 (16), 2216–2225.
(5) Sushkevich, V. L.; Palagin, D.; Ivanova, I. I. ACS Catal. 2015, 4833–4836.
(6) Taifan, W. E.; Bučko, T.; Baltrusaitis, J. J. Catal. 2017, 346, 78–91.
(7) Muller, P.; Burt, S. P.; Love, A. M.; McDermott, W. P.; Wolf, P.; Hermans, I. ACS
Catal. 2016, 6 (10), 6823–6832.
(8) Sushkevich, V. L.; Ivanova, I. I. Appl. Catal. B Environ. 2017, 215, 36–49.
(9) Chieregato, A.; Velasquez Ochoa, J.; Bandinelli, C.; Fornasari, G.; Cavani, F.;
Mella, M. ChemSusChem 2014, 8 (2), 377–388.
(10) Zhang, M.; Gao, M.; Chen, J.; Yu, Y. RSC Adv. 2015, 5 (33), 25959–25966.
(11) Taifan, W.; Yan, G. X.; Baltrusaitis, J. Catal. Sci. Technol. 2017, 7 (20), 4648–
4668.
(12) Fan, D.; Dong, X.; Yu, Y.; Zhang, M. Phys. Chem. Chem. Phys. 2017, 19 (37),
25671–25682.
(13) Hayashi, Y.; Akiyama, S.; Miyaji, A.; Sekiguchi, Y.; Sakamoto, Y.; Shiga, A.;
Koyama, T.; Motokura, K.; Baba, T. Phys. Chem. Chem. Phys. 2016, 18 (36),
25191–25209.
(14) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A.
ChemSusChem 2014, 7 (9), 2505–2515.
(15) Larina, O.; Kyriienko, P.; Soloviev, S. Catal. Letters 2015, 1–7.
(16) Da Ros, S.; Jones, M. D.; Mattia, D.; Pinto, J. C.; Schwaab, M.; Noronha, F. B.;
Kondrat, S. A.; Clarke, T. C.; Taylor, S. H. ChemCatChem 2016, 8, 1–12.
(17) Ohnishi, R.; Akimoto, T.; Tanabe, K. J. Chem. Soc. Chem. Commun. 1985, No.
22, 1613–1614.
(18) Sushkevich, V. L.; Ivanova, I. I.; Ordomsky, V. V; Taarning, E. ChemSusChem
2014, 7 (9), 2527–2536.
(19) Makshina, E. V.; Janssens, W.; Sels, B. F.; Jacobs, P. A. Catal. Today 2012, 198
(1), 338–344.
(20) Jones, M. D.; Keir, C. G.; Iulio, C. Di; Robertson, R. A. M.; Williams, C. V;
Apperley, D. C. Catal. Sci. Technol. 2011, 1 (2), 267–272.
(21) Bond, G. C.; Thompson, D. T. Catal. Rev. - Sci. Eng. 1999, 41 (3 & 4), 319–388.
(22) Guan, Y.; Hensen, E. J. M. J. M. Appl. Catal. A Gen. 2009, 361 (1), 49–56.
(23) Wittcoff, H. A. A. J. Chem. Educ. 1983, 60 (12), 1044.
(24) Sushkevich, V. L.; Ivanova, I. I.; Taarning, E. ChemCatChem 2013, 5 (8), 2367–
2373.
(25) Chang, F.-W.; Kuo, W.-Y.; Lee, K.-C. Appl. Catal. A Gen. 2003, 246 (2), 253–
264.
(26) Chang, F.-W.; Yang, H.-C.; Roselin, L. S.; Kuo, W.-Y. Appl. Catal. A Gen. 2006,
304 (0), 30–39.
267
(27) Janssens, W.; Makshina, E. V. V.; Vanelderen, P.; De Clippel, F.; Houthoofd, K.;
Kerkhofs, S.; Martens, J. A. A.; Jacobs, P. A. A.; Sels, B. F. B. F. ChemSusChem
2014, 8 (6), 994–1008.
(28) Angelici, C.; Meirer, F.; van der Eerden, A. M. J.; Schaink, H. L.; Goryachev, A.;
Hofmann, J. P.; Hensen, E. J. M.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS
Catal. 2015, 5 (10), 6005–6015.
(29) Kyriienko, P. I.; Larina, O. V; Soloviev, S. O.; Orlyk, S. M.; Calers, C.; Dzwigaj,
S.; Pisarzhevsky, L. V. ACS Sustain. Chem. Eng. 2017, 5 (3), 2075–2083.
(30) Sekiguchi, Y.; Akiyama, S.; Urakawa, W.; Koyama, T.; Miyaji, A.; Motokura, K.;
Baba, T. Catal. Commun. 2015, 68, 20–24.
(31) Baylon, R. a. L.; Sun, J.; Wang, Y. Catal. Today 2015, 1–7.
(32) Musolino, V.; Selloni, A.; Car, R. J. Chem. Phys. 1998, 108 (12), 5044.
(33) Pacchioni, G. J. Chem. Phys. 1996, 104 (18), 7329.
(34) Asakura, K.; Iwasawa, Y. Mater. Chem. Phys. 1988, 18, 499–512.
(35) Colonna, S.; Arciprete, F.; Balzarotti, A.; Fanfoni, M.; De Crescenzi, M.; Mobilio,
S. Surf. Sci. 2002, 512, L341–L345.
(36) Rodriguez, J. a.; Jirsak, T.; Chaturvedi, S. J. Chem. Phys. 1999, 111 (17), 8077.
(37) Rodriguez, J. a.; Jirsak, T.; Freitag, A.; Larese, J. Z. J. Phys. Chem. B 2000, 2
(100), 7439–7448.
(38) Zhukovskii, Y. F.; Kotomin, E. a.; Borstel, G. Vacuum 2004, 74 (2), 235–240.
(39) Matveev, A.; Neyman, K.; Yudanov, I.; Rösch, N. Surf. Sci. 1999, 426 (1), 123–
139.
(40) Galadima, A.; Muraza, O. J. Nat. Gas Sci. Eng. 2015, 25, 303–316.
(41) Zonetti, P. C. C.; Celnik, J.; Letichevsky, S.; Gaspar, A. B. B.; Appel, L. G. G. J.
Mol. Catal. A Chem. 2011, 334 (1–2), 29–34.
(42) Velu, S.; Suzuki, K.; Okazaki, M.; Kapoor, M. P.; Osaki, T.; Ohashi, F. J. Catal.
2000, 194 (2), 373–384.
(43) Ro, I.; Liu, Y.; Ball, M. R. R.; Jackson, D. H. K. H. K.; Chada, J. P. P.; Sener, C.;
Kuech, T. F. F.; Madon, R. J. J.; Huber, G. W. W.; Dumesic, J. A. A. ACS Catal.
2016, 6 (10), 7040–7050.
(44) Bravo-Suarez, J. J.; Subramaniam, B.; Chaudhari, R. V. J. Phys. Chem. C 2012,
116 (34), 18207–18221.
(45) Vlaic, G.; Bart, J. C. J. C. J.; Cavigiolo, W.; Mobilo, S. Chem. Phys. Lett. 1980, 76
(3), 453–459.
(46) Witzke, M. E. E.; Dietrich, P. J. J.; Ibrahim, M. Y. S. Y. S.; Al-Bardan, K.;
Triezenberg, M. D. D.; Flaherty, D. W. W. Chem. Commun. 2017, 53 (3), 597–
600.
(47) Chouillet, C.; Villain, F.; Kermarec, M.; Lauron-Pernot, H.; Louis, C. J. Phys.
Chem. B 2003, 107 (15), 3565–3575.
(48) Didenko, O. Z.; Kosmambetova, G. R.; Strizhak, P. E. Chinese J. Catal. 2008, 29
(11), 1079–1083.
(49) Pascual, J. L.; Savoini, B.; González, R. Phys. Rev. B 2004, 70 (4), 45109.
(50) Hosoglu, F.; Faye, J.; Mareseanu, K.; Tesquet, G.; Miquel, P.; Capron, M.;
Gardoll, O.; Lamonier, J. F.; Lamonier, C.; Dumeignil, F. Appl. Catal. A Gen.
2015, 504, 533–541.
(51) Laskowski, R.; Blaha, P.; Schwarz, K. Phys. Rev. B 2003, 67 (7), 75102.
268
(52) Breedon, M.; Spencer, M. J. S.; Yarovsky, I. Surf. Sci. 2009, 603 (24), 3389–3399.
(53) Gruver, V.; Sun, A.; Fripiat, J. J. Catal. Letters 1995, 34 (3–4), 359–364.
(54) Chung, S.-H.; Angelici, C.; Hinterding, S. O. M.; Weingarth, M.; Baldus, M.;
Houben, K.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS Catal. 2016, 6 (6),
4034–4045.
(55) Harvey, J. N. Kaltsoyannis, N., McGrady, J. E., Eds.; Springer Berlin Heidelberg:
Berlin, Heidelberg, Heidelberg, 2004; pp 151–184.
(56) Chizallet, C.; Costentin, G.; Che, M.; Delbecq, F.; Sautet, P.; Surface, D.; Marie,
P. J. Am. Chem. Soc. 2007, No. 19, 6442–6452.
(57) Gawande, M. B.; Goswami, A.; Felpin, F.-X.; Asefa, T.; Huang, X.; Silva, R.;
Zou, X.; Zboril, R.; Varma, R. S. Chem. Rev. 2016, 116 (6), 3722–3811.
(58) Marion, M. C.; Garbowski, E.; Primet, M. J. Chem. Soc.{,} Faraday Trans. 1990,
86 (17), 3027–3032.
(59) Coluccia, S.; Lavagnino, S.; Marchese, L. Mater. Chem. Phys. 1988, 18 (5), 445–
464.
(60) Yoshida, H.; Shimizu, T.; Murata, C.; Hattori, T. J. Catal. 2003, 220 (1), 226–232.
(61) Cavalleri, M.; Pelmenschikov, A.; Morosi, G.; Gamba, A.; Coluccia, S.; Martra, G.
In Oxide-based Systems at the Crossroads of ChemistrySecond International
Workshop October 8-11, 2000, Como, Italy; A. Gamba, C. C. and S. C. B. T.-S. in
S. S. and C., Ed.; Elsevier, 2001; Vol. Volume 140, pp 131–139.
(62) Valizadeh, H.; Azimi, A. A. J. Iran. Chem. Soc. 2011, 8 (1), 123–130.
(63) De Baerdemaeker, T.; Feyen, M.; Müller, U.; Yilmaz, B.; Xiao, F.-S.; Zhang, W.;
Yokoi, T.; Bao, X.; Gies, H.; De Vos, D. E. ACS Catal. 2015, 3393–3397.
(64) Ho, C. R.; Shylesh, S.; Bell, A. T. ACS Catal. 2016, 6 (2), 939–948.
(65) Liu, Q.; Liu, Z.; Li, C. Chinese J. Catal. 2006, 27 (7), 636–646.
(66) Díaz, G.; Pérez-Hernández, R.; Gómez-Cortés, A.; Benaissa, M.; Mariscal, R.;
Fierro, J. L. G. J. Catal. 1999, 187 (1), 1–14.
(67) Baltrusaitis, J.; Mendoza-Sanchez, B.; Fernandez, V.; Veenstra, R.; Dukstiene, N.;
Roberts, A.; Fairley, N. Appl. Surf. Sci. 2015, 326, 151–161.
(68) Georgieff, K. K. J. Appl. Polym. Sci. 1966, 10 (9), 1305–1313.
(69) Bevington, J. C.; Norrish, R. G. W. Proc. R. Soc. London. Ser. A. Math. Phys. Sci.
1949, 196 (1046), 363 LP-378.
(70) Song, H.; Ozkan, U. S. J. Catal. 2009, 261 (1), 66–74.
(71) Inui, K.; Kurabayashi, T.; Sato, S. Appl. Catal. A Gen. 2002, 237 (1–2), 53–61.
(72) Singh, M.; Zhou, N.; Paul, D. K.; Klabunde, K. J. J. Catal. 2008, 260 (2), 371–
379.
(73) Palagin, D.; Sushkevich, V. L.; Ivanova, I. I. J. Phys. Chem. C 2016, 120 (41),
23566–23575.
(74) Ordomsky, V. V.; Sushkevich, V. L.; Ivanova, I. I. J. Mol. Catal. A Chem. 2010,
333 (1), 85–93.
(75) Wulfers, M. J.; Tzolova-Müller, G.; Villegas, J. I.; Murzin, D. Y.; Jentoft, F. C. J.
Catal. 2012, 296, 132–142.
(76) Nordvang, E. C.; Borodina, E.; Ruiz-Martínez, J.; Fehrmann, R.; Weckhuysen, B.
M. Chem. – A Eur. J. 2015, 21 (48), 17324–17335.
(77) Kang, M.; Park, E. D.; Kim, J. M.; Yie, J. E. Catal. Today 2006, 111 (3–4), 236–
241.
269
(78) Xu, R.; Wei, W.; Li, W. H.; Hu, T. D.; Sun, Y. H. J. Mol. Catal. A Chem. 2005,
234 (1–2), 75–83.
(79) Choy, J.-H.; Kim, D.-K.; Hwang, S.-H.; Demazeau, G. Phys. Rev. B 1994, 50 (22),
16631–16639.
(80) Rehr, J. J.; Kas, J. J.; Vila, F. D.; Prange, M. P.; Jorissen, K. Phys. Chem. Chem.
Phys. 2010, 12 (21), 5503–5513.
(81) Velu, S.; Wang, L.; Okazaki, M.; Suzuki, K.; Tomura, S. Microporous
Mesoporous Mater. 2002, 54 (1–2), 113–126.
(82) Galoisy, L.; Cormier, L.; Calas, G.; Briois, V. J. Non. Cryst. Solids 2001, 293–295,
105–111.
(83) Wang, H.-C.; Wei, Y.-L.; Yang, Y.-W.; Lee, J.-F. J. Electron Spectros. Relat.
Phenomena 2005, 144–147, 817–819.
(84) Kelly, R. A.; Andrews, J. C.; DeWitt, J. G. Microchem. J. 2002, 71 (2), 231–245.
(85) Waychunas, G. A.; Fuller, C. C.; Davis, J. A.; Rehr, J. J. Geochim. Cosmochim.
Acta 2003, 67 (5), 1031–1043.
(86) Jeong, E.-S.; Park, C.; Jin, Z.; Yoo, J.; Yi, G.-C.; Han, S.-W. Journal of
Nanoscience and Nanotechnology. pp 1880–1883.
(87) Di Cosimo, J. I.; Dıez, V. K.; Xu, M.; Iglesia, E.; Apesteguıa, C. R. J. Catal. 1998,
178 (2), 499–510.
(88) Busca, G.; Lorenzelli, V. Mater. Chem. 1982, 7 (1), 89–126.
(89) T. Barzetti, E. Selli, D. Moscotti, L. F. J. Chem. Soc. Faraday Trans. 1996, 92 (8),
1401–1407.
(90) Lercher, J. A.; Grijndling, C.; Eder-Mirth, G. Catal. Today 1996, 27, 353–376.
(91) Lavalley, J. C. Catal. Today 1996, 27, 377–401.
(92) Burcham, L. J.; Briand, L. E.; Wachs, I. E. Langmuir 2001, 17 (20), 6164–6174.
(93) Badlani, M.; Wachs, I. E. Catal. Letters 2001, 75 (3–4), 137–149.
(94) Burcham, L. J.; Deo, G.; Gao, X.; Wachs, I. E. Top. Catal. 2000, 11–12, 85–100.
(95) Di Valentin, C.; Del Vitto, A.; Pacchioni, G.; Abbet, S.; Wörz, A. S. S.; Judai, K.;
Heiz, U. J. Phys. Chem. B 2002, 106 (46), 11961–11969.
(96) Kondo, J.; Sakata, Y.; Maruya, K.; Tamaru, K.; Onishi, T. Appl. Surf. Sci. 1987,
28, 475–478.
(97) Busca, G.; Lamotte, J.; Lavalley, J. ciaude; Lorenzelli, V. J. Am. Chem. Soc. 1987,
109 (17), 5197–5202.
(98) Kim, T.; Wachs, I. E. J. Catal. 2008, 255 (2), 197–205.
(99) Angelici, C.; Velthoen, M. E. Z.; Weckhuysen, B. M.; Bruijnincx, P. C. A. Catal.
Sci. Technol. 2015.
(100) Feng, T.; Vohs, J. M. J. Catal. 2002, 208 (2), 301–309.
(101) Redhead, P. a. Vacuum 1962, 12 (4), 203–211.
270
Chapter 7
Conclusions and future outlook
1. Conclusions ........................................................................................................270
2. Future Outlook ..................................................................................................273
References ......................................................................................................................275
1. Conclusions
Chapter 3
A complex reactive mechanism of ethanol to form 1,3-butadiene was explored
using periodic quantum chemical methods. Three reaction mechanisms, in particular, were
tested using DFT, namely Prins condensation, aldol condensation, and hemiacetal
rearrangements. Based on the thermodynamic and kinetic data determined within this study
we identified four rate important steps in the overall process, namely ethanol
dehydrogenation and dehydration to acetaldehyde and ethylene, respectively, aldol
condensation between adsorbed enolate and physisorbed acetaldehyde, and finally
carbanion stabilization between ethylene and acetaldehyde. In particular, ethanol
dehydration to form ethylene possessed lower energy barrier than dehydrogenation to yield
acetaldehyde suggesting competing reactive pathways. Aldol condensation step to form
acetaldol was preceded with forward free-energy barrier of 16.1 kcal/mol but limited
thermodynamically with endergonic reaction free energy of 12.9 kcal/mol. The energetic
barrier for the first C-C bond formation step in the Prins condensation mechanism at 28.8
kcal/mol also demonstrated the viability of this mechanism over an idealized MgO defect
sites. The model employed here represents a simplified, ideal defected MgO surface,
271
without taking into account participation of the formed Mg-O-Si bonds resulting from
interaction between MgO-SiO2 and OH groups, significantly affected the acidity-basicity
of the catalyst.
Chapter 4
Surface chemistry of WK (1:1) catalyst during the reaction of ethanol and the
corresponding reactive intermediates, including acetaldehyde, crotonaldehyde, crotyl
alcohol, was investigated using in situ DRIFTS measurements combined with DFT
calculations. Involvement of the native hydroxyl groups was shown to be transient, mostly
due to the hydrogen bonding with the intermediates. The stability of these OH groups
suggested that interaction between the catalyst and intermediates might be due to the
interaction between the Lewis metal heteroatom with the intermediates, instead of with the
OH groups. Ethanol adsorbed as both physisorbed and chemisorbed surface species, while
acetaldehyde, when formed exhibited high reactivity to yield crotonaldehyde but the excess
resulted in strongly bound surface species assigned to surface acetate, and/or 2,4-
hexadienal or polymerized acetaldehyde due to the basicity of the surface. Crotonaldehyde
was more likely to be reduced by ethanol to yield crotyl alcohol than desorbing, even at
relatively high temperatures. DRIFTS study of crotyl alcohol further elucidated the nature
of its interaction with the catalyst, where dissociative adsorption led to the deprotonation
of the molecule and C-O bond scission to yield 1,3-BD. Altogether, the data presented
unraveled a complex interplay between the surface hydroxyl groups, gaseous reactants and
surface bound reactive intermediates of 1,3-BD formation.
Chapter 5
MgO/SiO2 catalyst active surface sites were analyzed using in situ DRIFTS (using
272
complementary DFT calculations), TPRS and steady state reactor in combination with bulk
XRD and surface LEIS measurements. Combination of in-situ probing with CO2 and
pyridine and in-situ poisoning demonstrated the site requirement for the catalyst. In
particular, it was determined that the weak basic sites were responsible for ethanol
dehydrogenation, strong basic sites for aldol condensation and MPV reduction, while
stronger acid sites catalyzed acetaldol and crotyl alcohol dehydration reactions and weak
acid sites catalyzed the undesired ethanol dehydration. Furthermore, through a combination
of NH3-TPD and DFT the presence of open and closed LAS was identified while further
elaborating Mg coordination, as adopted from LAS classification of zeolitic materials.1–3
The MgSi-WK catalyst was shown to have both open LAS with both Mg3C and Mg4C as
the anchoring LAS, and a very isolated closed LAS (Mg3C).
Chapter 6
Cu- and Zn-promoted wet kneaded MgO/SiO2 catalysts were interrogated in situ
and operando and provided new insights into the structure and reactivity of their catalytic
sites during ethanol reaction to 1,3-BD. In-situ UV-Vis revealed the presence of Cu2+
species with dimeric coordination, while for Zn-promoted MgO/SiO2 catalyst, bulk ZnO
phase and Zn-MgO solid solution were observed. Promotion with metals showed increases
in weak basic sites, with Zn contributing to more Lewis acid sites that are responsible for
the enhanced activity. In-situ DRIFT spectroscopy results allowed decoupling of the aldol
condensation and dehydrogenation fundamental steps that took place on the surface
suggesting that promoting the catalyst with transition metal promoters enhanced the ability
of the catalyst to carry out aldol condensation as correlated with the steady state reactivity
experiments. In-situ UV-Vis spectroscopy suggested appearance of π-π* electronic
273
transitions of allylic cations, cyclic or aromatic species on the catalysts while also
providing insights on the oligomeric structure of the active sites. Our operando X-ray
measurements were combined with ab initio multiple scattering modelling to unravel the
exact electronic structure of the Cu and Zn promoters. Change in the local coordination of
Cu indicates the presence of more than one Cu2+ species, with only one contributes to the
reaction. This Cu2+ species was reduced to Cu0 during reaction via a Cu species that was
identified as a Cu species with reduced Cu-O coordination number. Zn-promotion, on the
other hand, resulted in a very stable catalyst with stable Zn local coordination, barely
changed during the reaction. However, this catalyst exhibited very high reactivity, which
results in the formation of carbonaceous deposit that further deactivated the catalyst.
2. Future Outlook
The most important issue in the Lebedev reaction is to design a selective catalyst
that possesses an optimum combination of redox, basic, and acid sites. The lack of suitable
spectroscopic methods hampers comprehensive characterization of MgO/SiO2 catalysts.
Up until now, structure-activity relationship has not been achieved yet, with previous
investigators can only indirectly correlate the amount of layered hydrous magnesium
silicate phase to the 1,3-BD yield.4 Work by Hayashi, et al. further proves that SiO2 is not
fundamentally required for this reaction, which suggests that more comprehensive
characterization is necessary to directly correlate the molecular structure of the catalyst that
actively catalyze the reaction.5 The presence of both open and closed Lewis acid sites
discovered in this work further open a new research pathway, where combination of
spectroscopic method and probe molecules is necessary to unravel their participation
274
during reaction. Further study to confirm the reaction mechanism is also important to
further elucidate the need of specific active sites. Understanding the mechanism of this
cascade reaction, combined with elucidation of the molecular structure of the catalyst will
lead to a more rational design of the catalyst. In particular, the presence of water and
acetaldehyde during the reaction needs to be investigated.6,7
This system is still on an early stage, with no consensus on which preparation
method, Mg/Si ratio, optimum transition metal loading, and calcination temperature
achieved. The synthesis parameter leads to different acidity and basicity of the catalyst,
which suggests the need to optimize these parameters. Optimized parameters will lead to a
superior catalyst with the best selectivity, and combined with knowledge of reaction
mechanism, a kinetic rate expression can be modeled to engineer an appropriate reactive
system.
275
References
(1) Harris, J. W.; Cordon, M. J.; Di Iorio, J. R.; Vega-Vila, J. C.; Ribeiro, F. H.;
Gounder, R. J. Catal. 2016, 335, 141–154.
(2) Boronat, M.; Concepcion, P.; Corma, A.; Navarro, M. T.; Renz, M.; Valencia, S.
Phys. Chem. Chem. Phys. 2009, 11 (16), 2876–2884.
(3) Boronat, M.; Concepción, P.; Corma, A.; Renz, M.; Valencia, S. J. Catal. 2005,
234 (1), 111–118.
(4) Chung, S.-H.; Angelici, C.; Hinterding, S. O. M.; Weingarth, M.; Baldus, M.;
Houben, K.; Weckhuysen, B. M.; Bruijnincx, P. C. A. ACS Catal. 2016, 6 (6),
4034–4045.
(5) Hayashi, Y.; Akiyama, S.; Miyaji, A.; Sekiguchi, Y.; Sakamoto, Y.; Shiga, A.;
Koyama, T.; Motokura, K.; Baba, T. Phys. Chem. Chem. Phys. 2016, 18 (36),
25191–25209.
(6) Velasquez Ochoa, J.; Bandinelli, C.; Vozniuk, O.; Chieregato, A.; Malmusi, A.;
Recchi, C.; Cavani, F. Green Chem. 2016, 18, 1653–1663.
(7) Zhu, Q.; Wang, B.; Tan, T. ACS Sustain. Chem. Eng. 2017, 5 (1), 722–733.
276
Curriculum Vitae
William E. Taifan
Address: 9 Duh Drive #131, Bethlehem, PA 18015
Phone: +1 510-219-0716; +62 812-1601-1180
Email: [email protected]
Place of Birth: Surabaya, Jawa Timur, Indonesia
Date of Birth: September 25, 1991
Parents: Fantasi Pola and Ernie Gonawan
Experience
Research Assistant at Lehigh University
August 2014 – present
Ethanol-to-chemicals catalysis research. Presently focused in n-butanol and 1,3-butadiene
synthesis, as well as ethanol as a hydrogenating agent for bio-oil. Side projects including
CO2 capture and theoretical work on oxidative coupling of methane (OCM) catalyst.
Intern at Chemisence, Inc.
February 2014 - June 2014 (5 months)
Research internship at an upcoming green tech startup in Silicon Valley. Research
focused on developing chemical sensors based on carbon black conductivity and affinity
with polymers.
Undergraduate Research Assistant at Biochemical Engineering Lab ITS
August 2012 - August 2013 (1 year 1 month)
Undergraduate research on separation and purification of an ester from a bio-oil.
Technical Expertise
• Matlab, Vienna Ab-initio Simulation Package (VASP) and Gaussian09
• Catalyst synthesis, testing and characterization
• Spectroscopy (IR, UV-Vis)
• GC, GC-MS
277
Academic Qualifications
Doctor of Philosophy (PhD) in Chemical Engineering
Lehigh University 2014-2018
Dissertation Title:
Advisor: Professor Jonas Baltrusaitis
Master of Science in Chemical Engineering
University of California at Berkeley 2013-2014
Bachelor of Engineering (with Honor)
Institut Teknologi Sepuluh Nopember Surabaya, Indonesia 2009-2013
Publications
“Operando structure determination of Cu and Zn on supported MgO/SiO2 catalysts
during ethanol conversion to 1,3-butadiene”
ACS Catalysis, submitted
Authors: William E. Taifan, Yuanyuan Li, John P. Baltrus, Anatoly I. Frenkel, Lihua
Zhang and Jonas Baltrusaitis
“In-situ spectroscopic insights on the molecular structure of the MgO/SiO2 catalytic
active site during ethanol conversion to 1,3-butadiene”
Journal of Physical Chemistry C, submitted
Authors: William Taifan and Jonas Baltrusaitis
“Surface chemistry of MgO/SiO2 catalysts during the ethanol catalytic conversion to
1,3-butadiene: in situ DRIFTS and DFT study”
Catalysis Science & Technology, 2017, 7(20), 4648-4668
Authors: William E. Taifan, George X. Yan, Jonas Baltrusaitis
“CH4 and H2S reforming to CH3SH and H2 catalyzed by metal promoted Mo6S8
cluster: a first-principles micro-kinetic study”
Catalysis Science & Technology, 2017, 7 (16), 3546-3554
Authors: William E. Taifan, Adam A. Arvidsson, Eric Nelson, Anders Hellman and
Jonas Baltrusaitis
“Minireview: direct catalytic conversion of sour natural gas (CH 4+ H 2 S+ CO 2)
components to high value chemicals and fuels”
Catalysis Science & Technology, 2017, 7 (14), 2919-2929
Authors: William E. Taifan and Jonas Baltrusaitis
“Catalytic conversion of ethanol to 1,3-butadiene on MgO: a comprehensive
mechanism elucidation using DFT calculations”
278
Journal of Catalysis, 2017, 346, 78–91
Authors: William E. Taifan, Tomas Bucko, Jonas Baltrusaitis
“Surface chemistry of carbon dioxide revisited”
Surface Science Reports, 2016, 71 (4), 595-671
Authors: William E. Taifan, Jean-Francois Boily, Jonas Baltrusaitis
“CH4 conversion to value added products: Potential, limitations and extensions of a
single step heterogeneous catalysis”
Applied Catalysis B: Environmental, 2016, 198, 525–547
Authors: William Taifan and Jonas Baltrusaitis
“Dairy wastewater for production of chelated biodegradable Zn micronutrient
fertilizers”
ACS Sustainable Chemistry & Engineering, 2016, 4 (3), 1722-1727
Authors: Hanyu Zhang, Megan Frey, Criztel Navizaga, Courtney Lenzo, Julian Taborda,
William Taifan, Abdolhamid Sadeghnejad, Alfredas Martynas Sviklas, Jonas Baltrusaitis
“Elucidation of ethanol to 1,3-butadiene reaction mechanism: Combined
experimental and DFT study”
Conference paper at Energy and Fuel – Biomass, ACS 2016 Fall Meeting at Philadelphia
Conference paper at Catalysis and Reaction Engineering Division, AIChE 2016 annual
meeting
Conference paper at North American Catalysis Society, 2017 annual meeting at Denver
Award and Honors
2016 Chevron Scholar Awards Lehigh Chemical Engineering
2017 Kokes Scholar Award – North American Catalysis Society
2017 John C. Chen Fellow – Lehigh University
Students Mentored (Past 4 Years)
• George X. Yan, B.S. Chemical Engineering, Lehigh University, Ph.D. Student, UCLA
• Paige Rockwell, B.S. Physics, Lycoming College
• Yiying Sheng, B.S. Chemical Engineering, Lehigh University